Dynamic adjustment method of PCS power in energy storage EMS system

Abstract

The present application relates to a method for dynamically adjusting the PCS power of an energy storage EMS system, which uses the Nth monitoring time node to the N+Mth monitoring time node as a demand calculation window, and uses a "slip" method and a weighted moving average algorithm to calculate the actual power demand of the starting monitoring time node of the next demand calculation window, that is, to calculate the actual power demand of the N+M+1th monitoring time node, sliding one monitoring time node each time. In addition, a PID control algorithm is used in the specific PCS power regulation. The EMS detects the power of the power grid meter in real time. Once it finds that it is about to exceed the demand, it will promptly control the PCS power meter and reduce the energy storage charging power, so as to achieve closed-loop control. It is not only accurate in the definition of the actual power demand, but also can control the PCS power according to the actual situation in the actual power control, and can quickly respond to changes in the energy storage EMS system and maintain system stability.

Description

Technical Field

[0001] The present application relates to the technical field of green and low-carbon industries, and in particular to a method for dynamically adjusting PCS power of an energy storage EMS system. Background Art

[0002] In my country's energy storage or industrial fields, industrial electricity is subject to a two-part electricity price, which consists of a basic electricity price and a kilowatt-hour electricity price. Figure 1 shown.

[0003] The basic electricity price refers to the electricity price calculated based on the capacity of the user's receiving transformer (kilovolt-amperes) or the maximum demand (kilowatts), and is the basic electricity fee for users to access the power grid; the kilowatt-hour electricity price is the electricity price calculated based on the time-of-use electricity price and the user's actual electricity consumption (kilowatts).

[0004] The State Grid collects users' electricity consumption data in real time, including voltage, power and other information, in a specific time period, usually 15 minutes, as the time window for calculating the current maximum demand. The average power consumption of users in the specified time period is calculated as the current maximum demand and used for basic electricity price settlement.

[0005] The difficulty in demand forecasting for EMS energy storage systems is mainly due to the large volatility of loads due to factors such as production plans, which makes demand forecasting more difficult. Demand forecasting relies on the analysis of historical data, but the incompleteness of the data will affect the accuracy of the forecast. A forecasting model needs to be established to consider the impact of various factors on demand. However, the current industrial and commercial energy storage EMS system lacks effective algorithms and models to more accurately predict demand.

[0006] Due to the above reasons, the current demand forecasting method of industrial and commercial energy storage EMS system is inaccurate, which may cause the user's maximum demand for the month to continue to increase, increasing the cost of the basic electricity price. Summary of the invention

[0007] Based on this, it is necessary to provide a dynamic adjustment method for the PCS power of an energy storage EMS system to address the problem that the traditional EMS energy storage system demand forecasting method predicts inaccurate demand values, which may lead to a continuous increase in the user's maximum demand for the month and increase the cost of the basic electricity price.

[0008] The present application provides a method for dynamically adjusting PCS power of an energy storage EMS system, including:

[0009] When the N+M+1th monitoring time node arrives, the average power of the power grid meter at each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node is obtained, and the estimated power demand at the N+M+1th monitoring time node is calculated according to the average power of the power grid meter at each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node using a weighted moving average algorithm; M is a positive integer with a fixed value, N is a positive integer whose value gradually increases by 1, and the initial value of N is 1,

[0010] Determine whether the estimated power demand at the N+M+1th monitoring time node is more consistent with the actual situation;

[0011] If the estimated power demand at the N+M+1th monitoring time node is more consistent with the actual situation, the estimated power demand at the N+M+1th monitoring time node is used as the actual power demand at the N+M+1th monitoring time node;

[0012] The PID algorithm is called to adjust the PCS power, and the real-time feedback power of the power grid meter is monitored in real time, so that the real-time feedback power of the power grid meter is less than or equal to the actual power demand at the N+M+1th monitoring time node, so as to complete the PCS power adjustment at the N+M+1th monitoring time node.

[0013] The present application relates to a method for dynamically adjusting the PCS power of an energy storage EMS system, which uses the Nth monitoring time node to the N+Mth monitoring time node as a demand calculation window, and uses a "slip" method and a weighted moving average algorithm to calculate the actual power demand of the starting monitoring time node of the next demand calculation window, that is, to calculate the actual power demand of the N+M+1th monitoring time node, sliding one monitoring time node each time. In addition, a PID control algorithm is used in the specific PCS power regulation. The energy storage EMS system detects the power of the power grid meter in real time. Once it finds that it is about to exceed the demand, it will adjust the PCS power in time to reduce the energy storage charging power, so as to achieve closed-loop control. It is not only accurate in the definition of the actual power demand, but also can control the PCS power according to the actual situation in the actual power control, and can quickly respond to changes in the energy storage EMS system and maintain system stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a diagram showing the basic structure of user electricity charges.

[0015] Figure 2 A method flow chart of a method for dynamically adjusting PCS power of an energy storage EMS system provided in one embodiment of the present application. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of this application more clear, the following is a further detailed description of this application in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain this application and are not used to limit this application.

[0017] The present application provides a method for dynamically adjusting PCS power of an energy storage EMS system.

[0018] like Figure 2 As shown, in one embodiment of the present application, the method for dynamically adjusting the PCS power of the energy storage EMS system includes:

[0019] S100, when the N+M+1th monitoring time node arrives, obtain the average power of the power grid meter at each monitoring time node between the Nth monitoring time node to the N+Mth monitoring time node, and use the weighted moving average algorithm to calculate the estimated power demand of the N+M+1th monitoring time node based on the average power of the power grid meter at each monitoring time node between the Nth monitoring time node to the N+Mth monitoring time node.

[0020] M is a positive integer with a fixed value, N is a positive integer whose value gradually increases by 1, and the initial value of N is 1.

[0021] S300: Determine whether the estimated power demand at the N+M+1th monitoring time node is more consistent with the actual situation.

[0022] S500: If the estimated power demand at the N+M+1th monitoring time node is more consistent with the actual situation, the estimated power demand at the N+M+1th monitoring time node is used as the actual power demand at the N+M+1th monitoring time node.

[0023] S700, call the PID algorithm to adjust the PCS power, and monitor the real-time feedback power of the power grid meter in real time, so that the real-time feedback power of the power grid meter is less than or equal to the actual power demand at the N+M+1th monitoring time node, so as to complete the PCS power adjustment at the N+M+1th monitoring time node.

[0024] Specifically, the average power of the power grid meter can be collected through the EMS data acquisition module. The EMS data acquisition module can achieve a sampling frequency of up to 200ms. Therefore, the time span between the Nth monitoring time node and the N+Mth monitoring time node can be 15 minutes. This period of time can be used as a demand calculation window. This embodiment uses the "slip" and weighted moving average algorithm to predict the actual power demand of the next monitoring time node.

[0025] For example, if the sampling frequency is 200ms / time, there are 60000ms in one minute, and M is 60000 divided by 200, which equals 300. The period from the 1st monitoring time node to the 301st monitoring time node is a demand calculation window, and a demand calculation window lasts for 1 minute. Then the estimated power demand at the 302nd monitoring time node is calculated based on the average power of the power grid meter at each monitoring time node between the 1st monitoring time node and the 301st monitoring time node, and then the actual power demand at the 302nd monitoring time node is determined based on whether the estimated power demand at the 302nd monitoring time node is more in line with the actual situation.

[0026] After obtaining the actual power demand at the 302nd monitoring time node, the demand calculation window is moved back by one monitoring time node, and the next demand calculation window is from the 2nd monitoring time node to the 302nd monitoring time node. The estimated power demand at the 303rd monitoring time node is calculated by the average power of the power grid meter at each monitoring time node between the 2nd monitoring time node and the 302nd monitoring time node. Then, the actual power demand at the 303rd monitoring time node is determined by whether the estimated power demand is more in line with the actual situation.

[0027] By analogy, starting from the 301st monitoring time node, each subsequent monitoring time node can obtain the corresponding actual power demand, which can be used as data support for power adjustment to ensure the accuracy of power adjustment.

[0028] The present application relates to a method for dynamically adjusting the PCS power of an energy storage EMS system, which uses the Nth monitoring time node to the N+Mth monitoring time node as a demand calculation window, and uses a "slip" method and a weighted moving average algorithm to calculate the actual power demand of the starting monitoring time node of the next demand calculation window, that is, to calculate the actual power demand of the N+M+1th monitoring time node, sliding one monitoring time node each time. In addition, a PID control algorithm is used in the specific PCS power regulation. The EMS detects the power of the power grid meter in real time. Once it finds that it is about to exceed the demand, it will promptly control the PCS power meter and reduce the energy storage charging power, so as to achieve closed-loop control. It is not only accurate in the definition of the actual power demand, but also can control the PCS power according to the actual situation in the actual power control, and can quickly respond to changes in the energy storage EMS system and maintain system stability.

[0029] In one embodiment of the present application, S100 includes, that is, when the N+M+1th monitoring time node arrives, obtaining the average power of the power grid meter of each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node, and using the weighted moving average algorithm to calculate the estimated power demand of the N+M+1th monitoring time node according to the average power of the power grid meter of each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node, including:

[0030] S110, when the N+M+1th monitoring time node arrives, the time between the Nth monitoring time node and the N+Mth monitoring time node is divided into a plurality of time periods with equal time spans.

[0031] S120, obtaining the grid meter power collected at each monitoring time node in each time period.

[0032] S130, calculating the arithmetic mean of the power grid meter power collected by all monitoring time nodes in each time period, and taking the arithmetic mean as the average power grid meter power of the time period, so as to obtain the average power grid meter power of each time period.

[0033] S140, obtaining the critical weight of each time period.

[0034] S150, calculating the weighted moving average value from the Nth monitoring time node to the N+Mth monitoring time node according to Formula 1.

[0035] Formula 1.

[0036] in, is the weighted moving value from the Nth monitoring time node to the N+Mth monitoring time node, i is the sequence number of the time period between the Nth monitoring time node and the N+Mth monitoring time node, is the average power of the power grid in the i-th time period between the N-th monitoring time node and the N+M-th monitoring time node, is the critical weight of the i-th time period between the N-th monitoring time node and the N+M-th monitoring time node, is the total number of time periods between the Nth monitoring time node and the N+Mth monitoring time node, is the symbol for the summation formula.

[0037] S150: Taking the weighted moving average value from the Nth monitoring time node to the N+Mth monitoring time node as the estimated power demand at the N+1th monitoring time node.

[0038] Specifically, S130 includes:

[0039] The average power of the power grid meter in the i-th time period between the N-th monitoring time node and the N+M-th monitoring time node is calculated according to formula 1-1.

[0040] Formula 1-1.

[0041] in, is the average power of the power grid in the i-th time period between the N-th monitoring time node and the N+M-th monitoring time node, L is the number of monitoring time nodes in the i-th time period between the N-th monitoring time node and the N+M-th monitoring time node, S is the sequence number of the monitoring time node, For the summation formula symbol, It is the grid meter power of the Sth monitoring time node in the ith time period between the Nth monitoring time node and the N+Mth monitoring time node.

[0042] Optionally, before calculating the weighted moving average value from the Nth monitoring time node to the N+Mth monitoring time node according to formula 1-1, the power of the power grid meter of each monitoring time node in the i-th time period between the Nth monitoring time node and the N+Mth monitoring time node is screened, and the specific screening process includes: removing the maximum power of the power grid meter, removing the minimum power of the power grid meter, and then calculating the average power of the power grid meter in the i-th time period between the Nth monitoring time node and the N+Mth monitoring time node according to formula 1-1. Therefore, L in formula 1-1 is reduced by 2 from the original value because two data points are missing.

[0043] Continuing with the example mentioned above, the sampling frequency is 200ms / time, there are 60000ms in one minute, M is 60000 divided by 200, which equals 300, and the period from the first monitoring time node to the 301st monitoring time node is a demand calculation window, and a demand calculation window lasts for 1 minute. Then Formula 2 calculates the arithmetic mean of the power of 301-2=299 grid meters.

[0044] In this embodiment, the arithmetic mean of the power of the power grid meters collected by all monitoring time nodes in each time period is calculated, which can ensure the averaging of the data and be more in line with the overall power situation of the time period. The subsequent introduction of weights for each time period can reflect the different importance of data in different time periods. The weighted moving average algorithm is combined to perform different degrees of smoothing and prediction to more accurately predict the demand value of the next time period, so that the final calculated estimated power demand for the N+M+1th monitoring time node is more in line with the actual power consumption.

[0045] In one embodiment of the present application, S140 includes, that is, obtaining the critical weight of each time period includes:

[0046] S141, obtaining the load power collected at each monitoring time node in each time period.

[0047] S142, calculating the arithmetic mean of the load powers collected by all monitoring time nodes in each time period, and taking the arithmetic mean as the load average power of the time period, so as to obtain the load average power of each time period.

[0048] S143, taking the maximum value of all load average powers as the load average power corresponding to the maximum critical weight, and taking the minimum value of all load average powers as the load average power corresponding to the minimum critical weight.

[0049] S144, normalize the load average power of each time period according to Formula 2 to map the load average power of each time period to a weight range of [minimum critical weight, maximum critical weight] to obtain the normalized load average power of each time period.

[0050] Formula 2.

[0051] in, is the normalized average load power of the i-th time period between the N-th monitoring time node and the N+M-th monitoring time node, i is the sequence number of the time period between the N-th monitoring time node and the N+M-th monitoring time node, is the average load power in the i-th time period between the N-th monitoring time node and the N+M-th monitoring time node, is the maximum value of the average power of all loads, It is the minimum value of the average power of all loads.

[0052] S145, using a linear interpolation method, converting the normalized load average power of each time period into a critical weight of each time period.

[0053] Specifically, linear mapping is a simple dynamic weight adjustment strategy, and this embodiment maps the average load power value to a predefined weight range. The load power can be mapped to the weight by linear interpolation according to the maximum and minimum power values ​​and the corresponding maximum and minimum weight values.

[0054] The maximum value of the critical weight can be set to 1, and the minimum value of the critical weight can be set to 0. Therefore, the critical weight range is [0, 1].

[0055] In one embodiment of the present application, S145 includes, that is, using the linear interpolation method to convert the normalized load average power of each time period into the critical weight of each time period, including:

[0056] S145a, converting the normalized load average power of each time period into the critical weight of each time period according to Formula 3.

[0057] Formula 3.

[0058] in, is the critical weight of the i-th time period between the N-th monitoring time node and the N+M-th monitoring time node, is the normalized average load power in the i-th time period between the N-th monitoring time node and the N+M-th monitoring time node, is the maximum value of critical weight, is the minimum critical weight.

[0059] Specifically, different time periods have different critical weights, and these critical weights assign different importance to the data in different time periods. Generally, the weight of the time period between the Nth monitoring time node and the N+Mth monitoring time node, which is far from the end of each time period, is greater, and the weight of the time period between the Nth monitoring time node and the N+Mth monitoring time node, which is far from the end of each time period, is smaller.

[0060] This is because the time period when the demand for the next monitoring cycle is about to be calculated is the most important, because the data jump is the smallest and the data is most consistent with the actual situation.

[0061] In one embodiment of the present application, before S100, the method further includes, that is, when the N+M+1th monitoring time node arrives, obtaining the average power of the power grid meter of each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node, and using a weighted moving average algorithm, before calculating the estimated power demand of the N+M+1th monitoring time node according to the average power of the power grid meter of each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node, the method further includes:

[0062] S010, obtaining the monthly planned power demand of this month, and taking the monthly planned power demand as the actual power demand of the N+Mth monitoring time node when N is 1.

[0063] Specifically, the historical data of the power grid meter and the historical data of the load meter in the latest Q years of the current system time are imported into the monthly power demand forecasting model database of the EMS station control cloud platform as the basic data for the monthly power demand forecasting. First, the basic data is analyzed, and the analysis of the basic data specifically includes:

[0064] S011, clarify the source of basic data and assign source labels to these basic data.

[0065] The sources of basic data may include the industrial park's electricity consumption information collection system, marketing business application system, smart meters and other channels.

[0066] S012, performing data cleaning on the basic data to which the data source label is attached, to obtain the basic data after data cleaning.

[0067] Data cleaning specifically includes checking whether the data is within a reasonable range, whether there are one or more outliers, and identifying and processing outliers according to actual conditions, such as eliminating or correcting them to ensure data accuracy.

[0068] S013, converting the basic data after data cleaning into a format suitable for database storage, and importing the basic data after data cleaning into a monthly planned power demand prediction model database using a database management tool or script.

[0069] Furthermore, the training of the monthly planned power demand prediction model is performed.

[0070] S014, creating a monthly planned power demand prediction model, dividing all data in the monthly planned power demand prediction model database into a training set and a test set in chronological order, and using the training set as training data to train the monthly planned power demand prediction model.

[0071] Specifically, the time series model ARIMA is selected, and the training set is used to train the monthly planned power demand prediction model.

[0072] S015, a cross-validation method is used to evaluate the prediction performance of the trained monthly planned power demand prediction model.

[0073] Specifically, the model parameters of the trained monthly planned power demand forecasting model are adjusted to maximize the forecasting accuracy.

[0074] S016, using the test set to evaluate the trained monthly planned power demand prediction model, comparing the difference between the prediction result and the actual value, evaluating the accuracy of the monthly planned power demand prediction model, and obtaining the tested monthly planned power demand prediction model.

[0075] S017, starting the monthly planned power demand forecasting model after the test, taking the total number of expected forecast months and standard data as input, and obtaining the forecast result output by the monthly planned power demand forecasting model after the test.

[0076] Specifically, the standard data is the national electricity consumption standard, and / or the provincial electricity consumption standard, and / or the municipal electricity consumption standard, and / or the district and county electricity consumption standard. For example, if the total number of months expected to be predicted is 12 months, then the monthly planned power demand for the next 12 months output by the tested monthly planned power demand prediction model will be obtained. The predicted 12-month planned power demand is sent to the EMS controller via MQTT, and the prediction effect of the tested monthly planned power demand prediction model is regularly monitored. The model parameters and features are adjusted according to the actual prediction result feedback to continuously optimize the prediction ability and stability of the monthly planned power demand prediction model.

[0077] The above weighted moving average algorithm is used to smooth the time series data to obtain the actual power demand at each monitoring time node, which is continuously sliding 24 hours a day until the beginning of the next month. The actual power demand at the last monitoring time node on the last day of the previous month is used as the monthly planned power demand for the next month.

[0078] In this embodiment, by introducing the monthly planned power demand and taking the monthly planned power demand as the actual power demand at the N+Mth monitoring time node when N is 1, the benchmark for adjusting the subsequent actual power demand can be anchored.

[0079] In one embodiment of the present application, S300 includes, that is, the determining whether the estimated power demand at the N+M+1th monitoring time node is more consistent with the actual situation, including:

[0080] S310 , determining whether the estimated power demand at the N+M+1th monitoring time node is greater than or equal to the actual power demand at the N+Mth monitoring time node.

[0081] S320: If the estimated power demand of the N+M+1th monitoring time node is greater than or equal to the actual power demand of the N+Mth monitoring time node, it is determined that the estimated power demand of the N+M+1th monitoring time node is more in line with the actual situation.

[0082] S330: If the estimated power demand at the N+M+1th monitoring time node is less than the actual power demand at the N+Mth monitoring time node, it is determined that the estimated power demand at the N+M+1th monitoring time node does not conform to the actual situation.

[0083] Specifically, in this embodiment, the actual power demand of the next monitoring time node is updated in real time, and the larger one is always selected between the estimated power demand of the next monitoring time node and the actual power demand of the previous monitoring time node as the actual power demand of the next monitoring time node.

[0084] In one embodiment of the present application, after S300, the method further includes, that is, after determining whether the estimated power demand at the N+M+1th monitoring time node is more in line with the actual situation, the method further includes:

[0085] S400: If the estimated power demand at the N+M+1th monitoring time node is not more consistent with the actual situation, the actual power demand at the N+Mth monitoring time node is used as the actual power demand at the N+M+1th monitoring time node.

[0086] Specifically, by selecting the larger one between the expected power demand at the next monitoring time node and the actual power demand at the previous monitoring time node as the actual power demand at the next monitoring time node, it can be ensured that each monitoring time node can update the power demand that best matches the actual situation in real time, making the system's demand forecast more accurate.

[0087] In one embodiment of the present application, S700 includes, that is, calling the PID algorithm to adjust the PCS power, and real-time monitoring the real-time feedback power of the power grid meter, so that the real-time feedback power of the power grid meter is less than or equal to the actual power demand at the N+M+1th monitoring time node, so as to complete the PCS power adjustment at the N+M+1th monitoring time node, including:

[0088] S710 , obtaining the power grid meter feedback power at the N+Mth monitoring time node, and determining whether the power grid meter feedback power is greater than the actual power demand at the N+M+1th monitoring time node.

[0089] S720, if the power grid meter feedback power at the N+Mth monitoring time node is less than or equal to the actual power demand at the N+M+1th monitoring time node, the PCS power is not adjusted, and the process returns to S100, that is, when the N+M+1th monitoring time node arrives, the average power of the power grid meter at each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node is obtained, and a weighted moving average algorithm is used to calculate the estimated power demand at the N+M+1th monitoring time node based on the average power grid meter at each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node.

[0090] Specifically, if the power is too low and does not exceed the actual power demand, no adjustment is required. Because the power intensity is too low, it can be understood that the power "can only decrease but not increase."

[0091] In one embodiment of the present application, S700 also includes, that is, calling the PID algorithm to adjust the PCS power, and real-time monitoring of the real-time feedback power of the power grid meter, so that the real-time feedback power of the power grid meter is less than or equal to the actual power demand at the N+M+1th monitoring time node, so as to complete the PCS power adjustment at the N+M+1th monitoring time node, and also includes:

[0092] S730: If the power grid meter feedback power at the N+Mth monitoring time node is greater than the actual power demand at the N+M+1th monitoring time node, the PCS power at the N+M+1th monitoring time node is calculated according to Formula 4.

[0093] Formula 4.

[0094] in, is the PCS power at time t, where t is the N+M+1th monitoring time node, is the proportional coefficient at time t, is the integral coefficient at time t, is the differential coefficient at time t, is the error at time t, is the integral of the historical error, is the derivative of the historical error.

[0095] Specifically, the PID algorithm is a classic control algorithm that is commonly used in industrial control systems. PID stands for proportional-integral-differential, which corresponds to the three control parameters in the algorithm. The PID controller calculates the output based on the current error (the difference between the set value and the actual value), the past error accumulation, and the error change rate. When the power exceeds the demand, the PID algorithm is used to adjust the behavior of the control system in a timely manner to make the actual value as close to the set value as possible. The PID algorithm is an effective and economical means of adjusting the power to the set target power.

[0096] PID power regulation is to quickly adjust to the target power, while demand prediction is to find the target power more accurately.

[0097] This embodiment introduces a PID control algorithm, which can provide stable control performance, quickly respond to system changes and maintain system stability.

[0098] In one embodiment of the present application, S730 includes, that is, calculating the PCS power of the N+M+1th monitoring time node according to Formula 4 includes:

[0099] S731, obtaining one or more of a system load change, a battery capacity, and a battery temperature curve at the N+M+1th monitoring time node.

[0100] S732, adjusting the proportional coefficient, integral coefficient, and differential coefficient of the N+M+1th monitoring time node by one or more of the system load change, battery capacity, and battery temperature curve at the N+M+1th monitoring time node.

[0101] Specifically, after S732, the following steps are performed:

[0102] S733, calculate the PCS power of the N+M+1th monitoring time node according to Formula 4.

[0103] Formula 4.

[0104] in, is the PCS power at time t, where t is the N+M+1th monitoring time node, is the proportional coefficient at time t, is the integral coefficient at time t, is the differential coefficient at time t, is the error at time t, is the integral of the historical error, is the derivative of the historical error.

[0105] Specifically, in the demand control of industrial and commercial energy storage for peak shaving and valley filling, it is very important to choose appropriate proportional coefficients, integral coefficients and differential coefficients, which will directly affect the performance and stability of the system.

[0106] 1) Proportional coefficient :

[0107] if If the setting is too small, the system response speed will be slow, which may cause the system to fail to quickly reach the set value; if If the value is set too large, the system may experience overshoot and oscillation.

[0108] 2) Integration coefficient :

[0109] if If the setting is too small, the system may not be able to completely eliminate the steady-state error; If the value is set too large, the system may experience overshoot and oscillation.

[0110] 3) Differential coefficient :

[0111] if If the setting is too small, the system may not be able to respond to rapidly changing errors in time; If the setting is too large, the system may become more sensitive to noise and interference, resulting in system instability.

[0112] In actual project applications, different projects have different hardware configurations. Usually, we first use automated simulation software to simulate and evaluate the impact of different parameter combinations on system performance in order to find the optimal control parameters as the factory default. , and Initial value.

[0113] The current maximum demand (KW) value based on the above prediction is used as the current maximum demand of the EMS system, that is, as the target value of the power grid table for adjusting the PCS power using the PID algorithm in the future.

[0114] In this embodiment, by adjusting appropriate parameters (proportional coefficient, integral coefficient, differential coefficient), the PID control algorithm can achieve good anti-reverse flow and capacity protection in the peak-shaving and valley-filling strategy of the power grid to achieve closed-loop control.

[0115] In practical applications, the factors that affect the PID control algorithm of the EMS system , and The factors mainly include the following:

[0116] 1. System load change

[0117] The EMS controller calculates the maximum load value minus the minimum load value once a minute as the load change within 1 minute, and then uses the weighted moving average algorithm to predict the load change in the next minute, with the weight increasing the closer to the current time period.

[0118] Rapid changes in system load may cause unstable system response, in which case the proportional coefficient needs to be increased. To enhance the sensitivity of the EMS controller to system changes, so as to quickly adjust the control output. At the same time, the continuous change of system load may lead to the accumulation of system deviation, in which case the integral coefficient needs to be increased. To increase the controller's ability to correct the system's steady-state error, so as to reduce the steady-state error. It may also cause system oscillation or overshoot, in which case the differential coefficient needs to be increased. To increase the controller's response speed to the system change rate and reduce oscillation and overshoot.

[0119] 2. Current battery capacity

[0120] Battery capacity may affect the charging and discharging speed of the system. When the energy storage system is charging, if the current battery capacity is low, it can be appropriately increased. , in order to speed up the charging response speed; if the current battery capacity is high, you can appropriately reduce To avoid overcharging. When the energy storage system is discharging, if the current battery capacity is high, you can increase the , in order to speed up the discharge response speed; if the current battery capacity is low, you can appropriately reduce to avoid over discharge.

[0121] 3. Battery temperature curve

[0122] The EMS controller records the average temperature of the battery cells in real time, and then uses a weighted moving average algorithm to predict the temperature value for the next minute, with the closer the time period is to the current one, the greater the weight.

[0123] When the battery temperature is high, the internal resistance of the battery will decrease, the charge and discharge rate may increase, and the system response speed may increase. , to avoid excessive system response and oscillation. When the battery temperature is low, the internal resistance of the battery will increase, the charge and discharge rate may slow down, and the system response speed may slow down. At this time, you can increase , to speed up system response speed and stability.

[0124] The EMS built-in adaptive PID control algorithm dynamically adjusts the PID algorithm according to system load changes, battery capacity SOC, and battery average temperature. , and Coefficients, of which three coefficients can be determined by Table 1. The influence ratio of each factor in different states of the system, while ensuring the algorithm , and The sum of the three coefficients is 1, which can dynamically adjust the PID coefficients to achieve the best control effect.

[0125] Table 1- Weight correspondence table of factors affecting PID coefficients

[0126]

[0127] The technical features of the above-described embodiments may be arbitrarily combined, and the execution order of the method steps is not limited. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0128] The above-described embodiments only express several real-time methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the present application. It should be noted that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for dynamically adjusting the PCS power of an energy storage EMS system, characterized in that: include: When the N+M+1th monitoring time node arrives, the average power of the power grid meter at each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node is obtained, and the estimated power demand at the N+M+1th monitoring time node is calculated according to the average power of the power grid meter at each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node using a weighted moving average algorithm; M is a positive integer with a fixed value, N is a positive integer whose value gradually increases by 1, and the initial value of N is 1, Determine whether the estimated power demand at the N+M+1th monitoring time node is more consistent with the actual situation; If the estimated power demand at the N+M+1th monitoring time node is more consistent with the actual situation, the estimated power demand at the N+M+1th monitoring time node is used as the actual power demand at the N+M+1th monitoring time node; Call the PID algorithm to adjust the PCS power, and monitor the real-time feedback power of the power grid meter in real time, so that the real-time feedback power of the power grid meter is less than or equal to the actual power demand at the N+M+1th monitoring time node, so as to complete the PCS power adjustment at the N+M+1th monitoring time node; The determining whether the estimated power demand at the N+M+1th monitoring time node is more consistent with the actual situation includes: Determine whether the estimated power demand at the N+M+1th monitoring time node is greater than or equal to the actual power demand at the N+Mth monitoring time node; If the predicted power demand at the N+M+1th monitoring time node is greater than or equal to the actual power demand at the N+Mth monitoring time node, it is determined that the predicted power demand at the N+M+1th monitoring time node is more consistent with the actual situation; If the predicted power demand at the N+M+1th monitoring time node is less than the actual power demand at the N+Mth monitoring time node, it is determined that the predicted power demand at the N+M+1th monitoring time node does not conform to the actual situation; After determining whether the estimated power demand at the N+M+1th monitoring time node is more consistent with the actual situation, the method further includes: If the estimated power demand at the N+M+1th monitoring time node is not more consistent with the actual situation, the actual power demand at the N+Mth monitoring time node is used as the actual power demand at the N+M+1th monitoring time node; The calling of the PID algorithm to adjust the PCS power and real-time monitoring of the real-time feedback power of the power grid meter makes the real-time feedback power of the power grid meter less than or equal to the actual power demand at the N+M+1th monitoring time node to complete the PCS power adjustment at the N+M+1th monitoring time node, including: Obtain the power feedback from the power grid meter at the N+Mth monitoring time node, and determine whether the power feedback from the power grid meter is greater than the actual power demand at the N+M+1th monitoring time node; If the power grid meter feedback power at the N+Mth monitoring time node is less than or equal to the actual power demand at the N+M+1th monitoring time node, the PCS power is not adjusted, and the return is made to the time when the N+M+1th monitoring time node arrives, and the average power grid meter power of each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node is obtained. The weighted moving average algorithm is used to calculate the estimated power demand of the N+M+1th monitoring time node based on the average power grid meter power of each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node.

2. The method for dynamically adjusting the PCS power of the energy storage EMS system according to claim 1 is characterized in that: When the N+M+1th monitoring time node arrives, the average power of the power grid meter of each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node is obtained, and the estimated power demand of the N+M+1th monitoring time node is calculated according to the average power of the power grid meter of each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node using a weighted moving average algorithm, including: When the N+M+1th monitoring time node arrives, the time between the Nth monitoring time node and the N+Mth monitoring time node is divided into multiple time periods with equal time spans; Obtain the grid meter power collected at each monitoring time node in each time period; Calculate the arithmetic mean of the power of the power grid collected by all monitoring time nodes in each time period, and use the arithmetic mean as the average power of the power grid in the time period to obtain the average power of the power grid in each time period; Get the critical weight of each time period; Calculate the weighted moving average from the Nth monitoring time node to the N+Mth monitoring time node according to Formula 1; Among them, WMA N～N+M is the weighted moving value from the Nth monitoring time node to the N+Mth monitoring time node, i is the sequence number of the time period between the Nth monitoring time node and the N+Mth monitoring time node, and X i is the average power of the grid meter in the i-th time period between the N-th monitoring time node and the N+M-th monitoring time node, W i is the critical weight of the i-th time period between the N-th monitoring time node and the N+M-th monitoring time node, K N～N+M is the total number of time periods between the Nth monitoring time node and the N+Mth monitoring time node, ∑ is the summation formula symbol; The weighted moving average value from the Nth monitoring time node to the N+Mth monitoring time node is used as the estimated power demand at the N+1th monitoring time node.

3. The method for dynamically adjusting the PCS power of the energy storage EMS system according to claim 2 is characterized in that: The step of obtaining the critical weight of each time period includes: Obtain the load power collected at each monitoring time node in each time period; Calculate the arithmetic mean of the load power collected by all monitoring time nodes in each time period, and use the arithmetic mean as the load average power of the time period to obtain the load average power of each time period; The maximum value of all load average powers is taken as the load average power corresponding to the maximum critical weight, and the minimum value of all load average powers is taken as the load average power corresponding to the minimum critical weight; The load average power of each time period is normalized according to Formula 2, so as to map the load average power of each time period to the weight range of [minimum critical weight, maximum critical weight], and obtain the normalized load average power of each time period; Among them, (normalized_power) i is the normalized average load power of the i-th time period between the N-th monitoring time node and the N+M-th monitoring time node, i is the sequence number of the time period between the N-th monitoring time node and the N+M-th monitoring time node, power i is the load average power in the i-th time period between the N-th monitoring time node and the N+M-th monitoring time node, max_power is the maximum value of all load average powers, and min_power is the minimum value of all load average powers; The normalized load average power of each time period is converted into the critical weight of each time period by using the linear interpolation method.

4. The method for dynamically adjusting the PCS power of the energy storage EMS system according to claim 3 is characterized in that: The method of using linear interpolation to convert the normalized load average power of each time period into the critical weight of each time period includes: According to Formula 3, the normalized load average power of each time period is converted into the critical weight of each time period; W i =(normalized_power) i ×(max_W-min_W)+min_W Formula 3; Among them, W i is the critical weight of the i-th time period between the N-th monitoring time node and the N+M-th monitoring time node, (normalized_power) i It is the normalized average load power in the ith time period between the Nth monitoring time node and the N+Mth monitoring time node, max_W is the maximum critical weight, and min_W is the minimum critical weight.

5. The method for dynamically adjusting the PCS power of the energy storage EMS system according to claim 4 is characterized in that: When the N+M+1th monitoring time node arrives, the average power of the power grid meter of each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node is obtained, and before the estimated power demand of the N+M+1th monitoring time node is calculated according to the average power of the power grid meter of each monitoring time node between the Nth monitoring time node and the N+Mth monitoring time node using a weighted moving average algorithm, the method further includes: The monthly planned power demand of this month is obtained, and the monthly planned power demand is used as the actual power demand of the N+Mth monitoring time node when N is 1.

6. The method for dynamically adjusting the PCS power of the energy storage EMS system according to claim 5, characterized in that: The calling of the PID algorithm to adjust the PCS power and real-time monitoring of the real-time feedback power of the power grid meter so that the real-time feedback power of the power grid meter is less than or equal to the actual power demand at the N+M+1th monitoring time node to complete the PCS power adjustment at the N+M+1th monitoring time node also includes: If the power grid meter feedback power at the N+Mth monitoring time node is greater than the actual power demand at the N+M+1th monitoring time node, the PCS power at the N+M+1th monitoring time node is calculated according to Formula 4; Among them, P t is the PCS power at time t, where t is the N+M+1th monitoring time node, is the proportional coefficient at time t, is the integral coefficient at time t, is the differential coefficient at time t, e(t) is the error at time t, is the integral of the historical error, is the derivative of the historical error.

7. The method for dynamically adjusting the PCS power of the energy storage EMS system according to claim 6, characterized in that: The PCS power of the N+M+1th monitoring time node is calculated according to Formula 4, including: Obtaining one or more of a system load change, a battery capacity, and a battery temperature curve at the N+M+1th monitoring time node; The proportional coefficient, integral coefficient and differential coefficient of the N+M+1th monitoring time node are adjusted by one or more of the system load change, battery capacity and battery temperature curve of the N+M+1th monitoring time node.

Images