A power utilization optimization control method and system for a heat exchange station power distribution system

Abstract

The present application relates to heat exchange station heat supply control and power distribution energy optimization technical field, especially in kind of heat exchange station power distribution system's electricity optimization control method and system. The method collects power distribution power, pump group parameters and secondary network temperature, pressure, flow data, predicts baseline heating power and electric load; The heat balance of pipe network and building thermal inertia is characterized, and the power distribution power flexible boundary is generated by combining the end pressure difference, pump group efficient area and electric-thermal-hydraulic sensitivity; The local sensitivity of pump frequency to electric power, heating power and end pressure difference is identified online; The pre-heating, peak clipping and back-up sequence is calculated, and the pre-heating and peak clipping process is constrained by the heat balance target; After the peak clipping forms heat shortage, according to the predicted total power, power distribution remaining capacity and online electric-thermal conversion coefficient, the back-up is staged. Thus, while meeting the heating and hydraulic constraints, invalid power transmission and distribution consumption and maximum demand are reduced, and the rebound after peak is inhibited.

Description

Technical Field

[0001] This invention relates to the fields of automatic control of centralized heating heat exchange stations, power distribution load management and demand-side power optimization technology, and in particular to a power optimization control method and system for the power distribution system of a heat exchange station. Background Technology

[0002] Centralized heating heat exchange stations are typically equipped with secondary network circulating pumps, makeup water pumps, primary side regulating mechanisms for heat exchangers, control cabinets, and several auxiliary electrical equipment. Among these, the circulating pumps operate continuously for extended periods, and their shaft power varies significantly with rotational speed, making them the main adjustable electrical load in the heat exchange station's power distribution system. Existing energy-saving controls for heat exchange stations often adjust pump frequency based on fixed time periods, outdoor temperature, secondary network pressure differential, supply and return water temperatures, or circulating pump efficiency curves. Some solutions also combine supply water temperature and flow rate adjustments based on predicted heat load.

[0003] The aforementioned control methods primarily determine "how much heat should be supplied at present" from the heating side or "how much pump frequency is needed at present" from the hydraulic side, while the power distribution side is usually only used as a passive power supply boundary. For heat exchange stations with basic electricity charges, maximum demand, grid demand response, or station capacity limitations, when multiple electrical devices increase their power at the same time, simply pursuing the minimum instantaneous heating error can easily lead to peak power demand. Conversely, if the pump frequency is forcibly reduced based solely on the peak power demand, it may result in insufficient terminal pressure differential, a drop in the temperature of heat users, and a concentrated increase in frequency to compensate after the peak shaving is completed, forming a new rebound peak.

[0004] On the other hand, the secondary network circulating water, heating pipe network, and building envelope have significant thermal inertia. The heating output and the user's thermal experience are not completely synchronized over a short period, giving some heating tasks time-transferability. However, existing heat exchange station controls typically do not translate this time-transferability into power flexibility boundaries that the power distribution system can directly use, defining "how many kilowatts can be increased, how many kilowatts can be decreased, and for how long." They also do not simultaneously identify the online impact of circulating pump frequency changes on electrical power, heating output, and terminal pressure differential.

[0005] Therefore, there is a need for a power consumption optimization control method that can unify thermal inertia, power demand constraints, and circulating pump hydraulic characteristics into the same control closed loop. This method would enable heat exchange stations to shift deferred heating tasks from peak power periods to off-peak or surplus power periods while meeting heating safety constraints, and control the post-peak replenishment process to reduce ineffective transmission and distribution power consumption and peak power distribution. Summary of the Invention

[0006] The purpose of this invention is to provide a power consumption optimization control method and system for the power distribution system of a heat exchange station, in order to solve the problems of the existing heat exchange station heating control and power distribution load optimization being disconnected, the difficulty in quantifying thermal inertia into adjustable electrical side capability, the tendency of peak shaving actions to cause heating constraint overstepping, and the formation of rebound peaks due to concentrated replenishment after peaks.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: First, power data acquisition terminals are installed on the main incoming line and major adjustable electrical branches of the heat exchange station. Temperature, pressure, and flow data acquisition terminals are installed on the supply and return water sides of the secondary network and on the most unfavorable loop to continuously obtain total active power, circulating pump branch power, circulating pump frequency, supply and return water temperature, flow rate, supply and return water pressure difference, and terminal pressure difference. Combined with outdoor weather forecasts and historical similar operating conditions, baseline heat load and baseline electrical load are formed for the future rolling control time domain.

[0008] Secondly, the secondary network circulating water, the heat transfer delay of the pipeline network, and the equivalent heat capacity of heat users are uniformly represented as a virtual thermal energy storage state. When the actual heat supply is higher than the baseline heat load, heat credit is generated; when the actual heat supply is lower than the baseline heat load, heat credit is consumed and heat deficit is generated. Heat credit or heat deficit is not directly used as the pump frequency setpoint, but is calculated together with the lower limit of terminal pressure difference, the supply and return water temperature range, the minimum frequency of the pump group, and the high-efficiency zone to form the power distribution flexibility boundary, thereby obtaining the power that can be increased or decreased and the sustainable duration in each future period.

[0009] Furthermore, by utilizing the minute changes in pump frequency naturally occurring during the operation of the heat exchange station or controlled identification actions, the local response of control quantity changes to total active power, actual heating power, and terminal pressure difference is calculated online, thus constructing an electro-thermal-hydraulic sensitivity matrix. This matrix is ​​used to answer the question "How much electrical power is reduced, how much heating power is lost, and how much terminal pressure difference is reduced by reducing the pump frequency by 1 Hz?", thereby enabling the peak shaving target on the electric side to be transformed into equipment actions that do not directly exceed hydraulic and thermal constraints.

[0010] In the rolling optimization layer, the optimization time domain is defined by multiple future control periods. It comprehensively considers time-of-use electricity prices, target demand lines, heat exchange station power distribution capacity, pump group high-efficiency zones, heating safety boundaries, remaining heat capacity, and equipment adjustment frequency to generate a continuous control sequence of pre-peak preheating, peak reduction, and post-peak replenishment. The optimization result is only applied to the first control period; then data is re-acquired and the solution is rolled over continuously to correct for weather forecast errors and user load variations.

[0011] In the fast execution layer, terminal pressure differential, total active power, and actual heating power are monitored on a shorter timescale than in the rolling optimization layer. If the actual state deviates from the rolling optimization prediction state, the minimum correction is calculated based on the electro-thermal-hydraulic sensitivity matrix, and the control quantity given by the rolling optimization is projected into the current safe and feasible domain. When the optimizer, communication, or critical sensors malfunction, the system switches to the original PLC backup control based on terminal pressure differential.

[0012] During the post-peak recovery phase, the circulating pumps are not allowed to directly return to the pre-peak frequency. Instead, the allowable increase in heat supply for each time period is determined based on the heat deficit, remaining power distribution capacity, and the shortest recovery time, allowing the deficit to gradually converge over multiple time periods. This avoids new demand peaks caused by multiple devices or heat exchange stations simultaneously increasing their frequency at the end of the demand response.

[0013] Preferably, the actual heating power is calculated according to Q_act(k)=rho*cp*q_v(k)*[T_s(k)-T_r(k)]; where rho is the density of the circulating medium, cp is the specific heat capacity of the circulating medium, q_v(k) is the volumetric flow rate of the secondary network, and T_s(k) and T_r(k) are the supply water temperature and return water temperature of the secondary network, respectively; when the unit of the flow rate collection value is m3 / h, it is divided by 3600 to convert it to m3 / s before being included in the calculation.

[0014] Preferably, the heat balance B(k+1) is defined as alpha_k*B(k)+[Q_act(k)-Q_base(k)]*Delta_t, where alpha_k=exp(-Delta_t / tau_th), tau_th is the equivalent heat time constant of the secondary network, heating pipe network and heat users, and B_min<=B(k)<=B_max is limited; B(k)>0 indicates heat credit, and B(k)<0 indicates heat debt.

[0015] Preferably, the electro-thermal-hydraulic sensitivity matrix is ​​updated online using a recursive least squares algorithm with a forgetting factor, and the operating conditions are identified according to the pump group combination status, valve opening range, or heating load range.

[0016] Preferably, the rolling optimization objective function is J=SUM{i=0...N-1}[w_E*c(k+i)*P(k+i)*Delta_t+w_B*[B(k+i)-B_ref(k+i)]^2+w_eff*phi_eff(k+i)+w_u*||Delta_u(k+i)||^2]+w_D*P_peak; where B_ref(k) is the heat balance target that changes with the pre-peak pre-heating, peak reduction and post-peak replenishment stages, taking a positive value before the peak, gradually decreasing during the peak, and approaching zero after the peak; P_peak satisfies P_peak>=P_bar_d(j), where P_bar_d(j) is the average power within the corresponding demand metering window.

[0017] Preferably, the thermal deficit D(k) = max[-B(k), 0] is defined, and the remaining distribution capacity P_spare(k) = max{P_lim(k) - P_pred^0(k) - P_safe, 0} is determined. For the compensation control direction Delta_u_r(k), when S_P(k)*Delta_u_r(k) > 0 and S_Q(k)*Delta_u_r(k) > 0, let kappa_k = [S_Q(k)*Delta_u_r(k)] / [S_P(k)*Delta_u_r(k)], and set 0 <= Q_repay(k) <= min{D(k) / tau_r, kappa_k*P_spare(k), Q_repay, max(k)}, so that the compensation process only occupies the current remaining distribution capacity, and is simultaneously limited by the maximum allowable compensation capacity on the thermal side.

[0018] Preferably, the control step size of the rolling optimization layer is 5 to 30 minutes, and the control step size of the fast execution layer is 10 to 120 seconds.

[0019] This invention has at least the following beneficial effects: 1. The thermal inertia is transformed from the "fuzzy buffer capability" on the heating side into an adjustable power boundary that can be directly used on the distribution side, so that the power demand control no longer depends on a fixed schedule or experience-based frequency reduction, and can clearly define the current power reduction and its duration.

[0020] 2. By using the electro-thermal-hydraulic sensitivity matrix to simultaneously correlate the reduction in electrical power with heat loss and changes in terminal pressure difference, the optimizer can prioritize actions with lower thermal and hydraulic costs per unit of electrical power reduction, thereby reducing ineffective hydraulic transport and avoiding terminal pressure loss caused by simple electrical peak shaving.

[0021] 3. Pre-peak heat charging, peak reduction, and post-peak heat deficit repayment share the same heat balance status; forward heating provides available heat credit for subsequent peak reduction, and the deficit generated by peak reduction directly constrains the repayment speed. The three-stage closed-loop coupling can suppress the synchronous frequency rebound after the end of demand response.

[0022] 4. Rolling optimization and fast constraint projection use different time scales. The former is responsible for economic and demand targets, while the latter is responsible for real-time hydraulic safety. The combination of the two can keep the control strategy executable when there are errors in the prediction, and retain the original PLC differential pressure closed loop as a safe degradation path.

[0023] 5. Compared with peak shifting using only thermal inertia, this scheme simultaneously limits the high-efficiency zone of the circulating pump and the hydraulic safety boundary, which can avoid the pump set entering a low-efficiency operating condition for preheating; compared with peak shaving only in power distribution, it also uses thermal credit to expand the shaving space, thus simultaneously improving peak demand and transmission efficiency, forming a synergistic technical effect of mutual gain between the electrical and thermal sides. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments are briefly described below.

[0025] Figure 1 This is a schematic diagram of the overall connection of the power distribution and heating control system for the heat exchange station of the present invention; Figure 2 This is a flowchart of the power consumption optimization control method of the present invention; Figure 3 This is a schematic diagram illustrating the dual-timescale control of rolling optimization, rapid correction, and hot debt management in this invention.

[0026] In the diagram: 1. Power grid incoming line; 2. Distribution transformer or main distribution cabinet; 3. Station distribution busbar; 4. Circulating pump frequency converter; 5. Makeup water pump and auxiliary load branch; 6. Circulating pump; 7. Makeup water pump and auxiliary equipment; 8. Heat exchanger; 9. Secondary network; 10. Heat user; 11. Data acquisition module; 12. Optimization controller; 13. PLC and control execution module; 201. Baseline prediction unit; 202. Heat flexible mapping unit; 203. Rolling optimization unit; 204. Real-time data feedback unit; 205. Sensitivity correction unit; 206. Control execution unit; 207. Heat balance and heat debt management unit. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the described embodiments are for explaining the invention and not for limiting the invention.

[0028] Reference Figure 1The heat exchange station is connected to the station's power distribution bus 3 via the power grid incoming line 1, a distribution transformer, or the main distribution cabinet 2. The power distribution bus 3 supplies power to the circulating pump frequency converter 4, the makeup water pump, and the auxiliary load branch 5. The circulating pump frequency converter 4 drives the circulating pump 6, and the makeup water pump and auxiliary load branch 5 correspondingly drive the makeup water pump and auxiliary equipment 7. On the heating side, heat is transferred from the heat exchanger 8 to the secondary network 9, which then delivers the heat to the heat users 10. The data acquisition module 11 simultaneously collects data from both the power distribution side and the heating side and provides real-time status information to the optimization controller 12. After calculating the target control quantity, the optimization controller 12 sends it to the circulating pump frequency converter 4 or other adjustable loads via the PLC and control execution module 13.

[0029] In this embodiment, the sampling period of the main incoming power acquisition terminal can be 1 to 10 seconds, and the sampling period of the supply and return water temperature, flow rate, and pressure data can be 5 to 60 seconds. To reduce the impact of instantaneous noise, the optimization controller can perform median filtering, amplitude limiting, and time alignment on the raw data, and fill in missing data by weighting the most recent valid value with the predicted value.

[0030] In step S1, the baseline is not a fixed rated power, but rather describes "the state to which the heat exchange station should naturally operate under the current meteorological and heat load conditions if power optimization is not performed." An initial baseline can be obtained by selecting a historical window with similar outdoor temperature, wind speed, time of day, and weekday attributes from the most recent 30-60 days of operational data, and then corrected based on the current representative room temperature deviation. For sites with machine learning capabilities, gradient boosting trees, temporal convolutional networks, or recurrent neural networks can also be used to predict the future baseline, but the prediction algorithm itself does not constitute a necessary limitation of this invention.

[0031] In step S2, the concept of virtual thermal energy storage is used to characterize the transferable heating task. Specifically, the actual heating power Q_act(k) = rho*cp*q_v(k)*[T_s(k)-T_r(k)]; when the flow rate is collected in m3 / h, it is first divided by 3600 to convert it to m3 / s. The heat balance is recursively calculated according to B(k+1) = alpha_k*B(k) + [Q_act(k)-Q_base(k)]*Delta_t, where alpha_k = exp(-Delta_t / tau_th), tau_th is the equivalent thermal time constant of the secondary network, heating pipe network and heat users, and B_min <= B(k) <= B_max is set. When Q_act(k) is higher than Q_base(k), heat credit B(k) > 0 is formed under the condition that the supply water temperature, representative room temperature and terminal pressure difference all meet the constraints; when Q_act(k) is lower than Q_base(k), heat credit is consumed and heat debt B(k) < 0 may be formed.

[0032] The conversion from heat balance to the electrical power flexible boundary does not use a fixed conversion factor, but rather combines the local sensitivity obtained in step S3. Let the control vector u include the frequency of each circulating pump, the pump group start / stop status, and the selectable auxiliary adjustable load status; near the current operating condition, Delta_P(k) = S_P(k) * Delta_u(k), Delta_Q(k) = S_Q(k) * Delta_u(k), Delta_p_end(k) = S_p(k) * Delta_u(k). The power threshold for downward adjustment is calculated using Delta_P_down,max(k) = max{-S_P(k)*Delta_u(k)}, and the power threshold for upward adjustment is calculated using Delta_P_up,max(k) = max{S_P(k)*Delta_u(k)}. The solution must satisfy at least B_min <= alpha_k*B(k) + [Q_act(k) + S_Q(k)*Delta_u(k) - Q_base(k)]*Delta_t<=B_max, p_end(k)+S_p(k)*Delta_u(k)>=p_min, u_min<=u(k)+Delta_u(k)<=u_max, |Delta_u(k)|<=r_u,max, and P_pred^0(k)+S_P(k)*Delta_u(k)<=P_lim(k)-P_safe. This maps the heat balance, hydraulic safety, and power distribution capacity together to the adjustable power boundaries for the current time period.

[0033] In step S3, the sensitivity matrix is ​​updated online. A recursive least squares algorithm with a forgetting factor is used, with the control variable Delta_u(k) as the regression vector varphi(k), and Delta_P(k), Delta_Q(k), and Delta_p_end(k) as the outputs z(k). The recursive gain is K(k) = C(k-1)*varphi(k) / [lambda + varphi(k)^T*C(k-1)*varphi(k)]; the parameter update is theta_hat(k) = theta_hat(k-1) + K(k)*[z(k) - varphi(k)^T*theta_hat(k-1)]; and the covariance update is C(k) = [C(k-1) - K(k)*varphi(k)^T*C(k-1)] / lambda. By identifying the changes in total active power, actual heating power, and terminal pressure difference, S_P(k), S_Q(k), and S_p(k) are obtained. Among them, C(k) is specifically used to represent the parameter estimation covariance matrix to avoid sign conflicts with the active power P(k) of the heat exchange station.

[0034] To avoid distortion of local models due to large-scale changes in operating conditions, multiple operating condition clusters can be established according to pump group combinations, secondary network flow ranges, or outdoor temperature ranges. When the optimization controller enters a new operating condition cluster, it uses the corresponding historical sensitivity as the initial value and continues to correct it with real-time data. If the parameter residuals continue to exceed the threshold, the confidence level of the matrix is ​​reduced, and the available flexibility boundary is narrowed until sufficient new data is obtained.

[0035] In step S4, refer to Figure 2 and Figure 3 The rolling optimization unit 203 establishes an optimization problem for the next N control periods during each calculation. If the control step size is 15 min and the prediction time domain is 4 h, then N is 16. The objective function can be written as J = SUM{i=0...N-1}[w_E*c(k+i)*P(k+i)*Delta_t+w_B*[B(k+i)-B_ref(k+i)]^2+w_eff*phi_eff(k+i)+w_u*||Delta_u(k+i)||^2]+w_D*P_peak. Where B_ref(k) is the heat balance target: set to a positive value before the peak to encourage the formation of heat credit, gradually decreasing during the peak period, and approaching zero after the peak; P_peak satisfies P_peak>=P_bar_d(j), where P_bar_d(j) is the average power within the demand metering window j. Therefore, the objective function optimizes the maximum demand in a billing sense directly, while allowing pre-peak heat charging, rather than just penalizing instantaneous sampling power.

[0036] When a future power distribution peak is predicted, the optimizer can slightly increase Q_act(t) above Q_base(t) during the pre-peak capacity surplus period to create heat credit, while requiring the pumps to remain in the high-efficiency zone to avoid inefficient preheating caused by simply increasing the frequency. After entering the peak period, the optimizer reduces the circulation pump frequency based on the heat credit and the terminal pressure differential safety margin, thus reducing electrical power; when the heat credit is exhausted or the representative room temperature approaches the lower limit, it automatically stops further reduction.

[0037] Step S5 employs a rapid correction. If the rolling optimization gives a target frequency of 35Hz for the circulating pump, and the real-time terminal pressure difference is already close to the lower limit, then the rapid correction unit 205 estimates the minimum compensation frequency based on S_p, and projects the corrected target frequency onto a feasible region where the terminal pressure difference, maximum frequency change rate, and power distribution are all satisfied, while maintaining a frequency "as close as possible to 35Hz". Conversely, if the real-time terminal pressure difference margin is greater than the predicted value and the total power is close to the demand line, it can be further reduced within the flexible boundary.

[0038] In step S6, when the peak segment Q_act(k) < Q_base(k), B(k) may become a negative value, and D(k) = max[-B(k), 0] is defined as the heat deficit. First, the predicted total active power P_pred^0(k) of the heat exchange station when no additional replenishment action is performed is calculated, which includes the predicted normal operating power of the current circulating pump, make-up pump, auxiliary equipment and non-adjustable loads, and then P_spare(k) = max{P_lim(k)-P_pred^0(k)-P_safe, 0} is determined. For the selected replenishment control direction Delta_u_r(k), when S_P(k)*Delta_u_r(k) > 0 and S_Q(k)*Delta_u_r(k) > 0, let kappa_k = [S_Q(k)*Delta_u_r(k)] / [S_P(k)*Delta_u_r(k)], and limit the replenishment power of each time period according to 0 <= Q_repay(k) <= min{D(k) / tau_r, kappa_k*P_spare(k), Q_repay, max(k)}. Thus, the replenishment capacity is determined by the actual remaining power distribution capacity and the current online sensitivity, which avoids incorrectly counting the power already occupied by the circulating pump into the surplus capacity.

[0039] For example, it is predicted that a certain station will have a demand peak from 17:00 to 19:00, and the power distribution margin is relatively large from 16:00 to 17:00. The optimizer can increase the heat supply of the secondary network to 1.03 to 1.08 times of the baseline from 16:00 to 17:00, so that the heat balance enters a positive range; from 17:00 to 19:00, the frequency of the circulating pump is reduced under the condition that the end pressure difference and the lower limit of the representative room temperature are satisfied; after 19:00, if other loads in the station are still high, operate with a small replenishment amount first, and gradually increase the replenishment amount after the total power drops. The above ratios are only used to illustrate the control logic, and the actual values are obtained by rolling optimization according to the constraints of equipment and heat users.

[0040] Reference Figure 3 , the baseline prediction unit 201, the heat flexibility mapping unit 202 and the rolling optimization unit 203 form a planning layer with a slower time scale; the real-time data feedback unit 204, the sensitivity correction unit 205 and the control execution unit 206 form an execution layer with a faster time scale; the heat balance and heat deficit management unit 207 provides status to both layers at the same time, so that the planning layer knows the remaining peak-shifting capacity, and the execution layer knows when to stop curtailment or start replenishment.

[0041] This invention also provides a corresponding power optimization control system. The system can be deployed on an industrial computer or PLC host computer in a single heat exchange station, or it can be deployed on an edge server across multiple heat exchange stations. For multi-station scenarios, the power flexibility boundary of each station can be calculated independently first, and then the demand target of each station can be allocated by the upper-level scheduler; the local station optimization controller still ensures local heating security according to the heat balance, sensitivity, and heat deficit rules of this invention.

[0042] When the total incoming power, terminal differential pressure, or key temperature sensor has no valid value for several consecutive sampling cycles, or when the optimization solution has no feasible solution within the specified time, the control execution module 13 stops using the optimization result and switches to the original terminal differential pressure or water supply temperature closed-loop control. After the key data is recovered and stabilized for several consecutive cycles, the optimization control is gradually restored to avoid sudden changes in pump frequency caused by the switch of control.

[0043] As can be seen from the above embodiments, this invention does not simply list electricity price optimization, thermal inertia, and water pump frequency regulation side by side. Instead, it utilizes the heat balance to generate a flexible boundary on the electricity side, uses electro-thermal-hydraulic sensitivity to map the electricity side target into the action of safety equipment, and then uses the thermal deficit to feed back the peak shaving result to the subsequent power compensation limit. The output of the previous stage directly constitutes the constraint or state of the next stage, thus forming a closed-loop coordination.

[0044] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments, and various changes and modifications can be made without departing from the spirit and scope of the invention; all such changes and modifications fall within the scope of protection claimed by the present invention.

Claims

1. A method for optimizing power consumption control in a heat exchange station power distribution system, characterized in that, Includes the following steps: S1. Obtain the total active power of the power distribution system of the heat exchange station, the branch power of each adjustable electrical equipment, and the supply water temperature, return water temperature, flow rate, supply and return water pressure difference, terminal pressure difference and circulation pump operating frequency of the secondary network of the heat exchange station. Based on historical operating data, outdoor meteorological data and current operating data, predict the baseline heat load and baseline electrical load in the rolling control time domain. S2. Based on the heat capacity of the secondary network circulating water, the pipeline transmission delay, and the equivalent heat capacity of the heat users, determine the available heat balance relative to the minimum heating constraint. Convert the available heat balance into the heating power that can be moved forward or delayed in each control period. Combined with the lower limit of terminal pressure difference, the supply and return water temperature boundaries, the high-efficiency operation zone of the circulating pump, and the electro-thermal-hydraulic sensitivity matrix, obtain the upward and downward power adjustment boundaries corresponding to the power distribution system of the heat exchange station. S3. Based on the changes in the frequency of the circulating pump or the combination state of the pump group during the continuous control period, as well as the corresponding changes in total active power, actual heating power and terminal pressure difference, identify the electric-thermal-hydraulic sensitivity matrix online. The electric-thermal-hydraulic sensitivity matrix includes at least the sensitivity of electric power to the control quantity, the sensitivity of heating power to the control quantity and the sensitivity of terminal pressure difference to the control quantity. S4. Within the rolling control time domain, with the goal of minimizing the weighted sum of heat exchange station electricity costs, maximum demand deviation, heat balance deviation, penalty amount for circulating pumps deviating from the high-efficiency operating zone, and equipment adjustment changes, within the above power adjustment boundary, lower power adjustment boundary, power distribution capacity boundary, and heating safety boundary, the equipment control sequence including the pre-peak pre-heating stage, peak reduction stage, and post-peak replenishment stage is obtained. S5. Send the first control quantity of the equipment control sequence to the circulating pump frequency converter and / or the adjustable electrical equipment in the station, and perform constraint projection correction on the first control quantity using the electro-thermal-hydraulic sensitivity matrix based on the deviation between the real-time terminal pressure difference and the real-time heating power and the predicted value. S6. Update the heat balance based on the difference between the baseline heating power and the actual heating power. When the heat balance is less than zero, it is recorded as a heat deficit. Determine the upper limit of the replenishment based on the remaining power distribution capacity in the next control period, the current local conversion coefficient of electric heating and the heat deficit, so that the post-peak replenishment power does not exceed the power distribution capacity boundary and the heat deficit gradually converges. S7. Repeat steps S1 to S6 to form a rolling power consumption optimization control for the power distribution load of the heat exchange station.

2. The power consumption optimization control method according to claim 1, characterized in that, In step S1, the baseline heat load is expressed as the baseline heating power and determined according to Q_base(k)=rho*cp*q_v,base(k)*[T_s,base(k)-T_r,base(k)]; where rho is the density of the circulating medium, cp is the specific heat capacity of the circulating medium, q_v,base(k) is the baseline volumetric flow rate, and T_s,base(k) and T_r,base(k) are the baseline supply water temperature and the baseline return water temperature, respectively; when the flow rate unit output by the flow acquisition terminal is m3 / h, it is divided by 3600 to convert it to m3 / s before being included in the calculation; the baseline volumetric flow rate and the baseline supply and return water temperatures are corrected based on historical similar weather days, outdoor temperature prediction values, and representative room temperature deviations.

3. The power consumption optimization control method according to claim 1, characterized in that, In step S2, the secondary network circulating water, heating pipe network, and heat users are equivalent to virtual thermal energy storage units. The heat balance B(k+1) is defined as alpha_k*B(k)+[Q_act(k)-Q_base(k)]*Delta_t, where alpha_k=exp(-Delta_t / tau_th), Q_act(k) is the actual heating power, Q_base(k) is the baseline heating power, Delta_t is the control step size, and tau_th is the equivalent thermal time constant corresponding to the secondary network, heating pipe network, and heat users. B_min<=B(k)<=B_max is also defined. When B(k)>0, it indicates that there is heat credit that can be released in subsequent control periods. When B(k)<0, it indicates that there is a heat deficit to be replenished. Based on the margin of the heat balance from B_min and B_max, and combined with the electro-thermal-hydraulic sensitivity matrix, the upward and downward power adjustment boundaries for the current control period are determined.

4. The power consumption optimization control method according to claim 1, characterized in that, In step S3, the recursive least squares algorithm with a forgetting factor is used to update the electro-thermal-hydraulic sensitivity matrix online. The change in control quantity Delta_u(k) is used as the regression vector varphi(k), and the change in total active power Delta_P(k), the change in actual heating power Delta_Q(k), and the change in terminal pressure Delta_p_end(k) are used as the output quantities z(k), according to K(k)=C(k-1)*varphi(k) / [lambda+varphi(k)^T*C(k-1)*varphi(k)], theta_hat(k)=theta_hat(k-1)+K(k)*[z(k)-varphi(k)]. The parameters are updated using phi(k)^T*theta_hat(k-1)] and C(k)=[C(k-1)-K(k)*varphi(k)^T*C(k-1)] / lambda; where K(k) is the recursive gain, C(k) is the parameter estimation covariance matrix, lambda is the forgetting factor, and theta_hat(k) is the sensitivity parameter estimate; the recursive identification is performed on Delta_P(k), Delta_Q(k), and Delta_p_end(k) respectively to obtain S_P(k), S_Q(k), and S_p(k); when the starting and stopping of the circulating pump, large valve movement, or heating conditions crossing the preset range are detected, an independent identification window is set for the corresponding conditions.

5. The power consumption optimization control method according to claim 1, characterized in that, In step S4, a heat balance target B_ref(k) that varies with the control phase is set. Before the predicted power distribution peak, B_ref(k) is set to > 0 to form a pre-charge heat target. During the power distribution peak, B_ref(k) is gradually reduced, and during the post-peak replenishment phase, B_ref(k) is brought close to zero. The rolling optimization objective function includes at least the electricity cost, maximum demand, heat balance tracking deviation, circulating pump efficiency loss, and control quantity change penalty. The maximum demand is determined according to the average power within the demand metering window. When the pump group has discrete start-stop states, mixed integer optimization is used to solve the problem. When only the frequency of the variable frequency is adjusted, constrained quadratic programming is used to solve the problem.

6. The power consumption optimization control method according to claim 1, characterized in that, In step S6, the heat deficit D(k) = max[-B(k), 0] is defined, and the remaining power distribution capacity P_spare(k) = max{P_lim(k) - P_pred^0(k) - P_safe, 0} is determined based on the predicted total active power of the heat exchange station without additional compensation actions. Here, P_pred^0(k) includes the predicted total power generated by the current circulating pump, makeup water pump, auxiliary equipment, and non-adjustable loads under normal operating conditions. For the selected compensation control direction Delta_u_r(k), when S_P(k) * Delta_u_r(k) > 0 and S_Q(k) * Delta_u_r(k) > 0, the remaining power distribution capacity is determined. When k)>0, determine the local electrothermal conversion coefficient kappa_k=[S_Q(k)*Delta_u_r(k)] / [S_P(k)*Delta_u_r(k)] under the current operating condition, and limit the post-peak recovery power according to 0<=Q_repay(k)<=min{D(k) / tau_r,kappa_k*P_spare(k),Q_repay,max(k)}; where tau_r is the minimum allowable recovery time, and Q_repay,max(k) is the maximum allowable recovery power on the hot side determined according to the upper limit of the water supply temperature, the terminal pressure difference, the allowable frequency of the pump group, and the high-efficiency operating zone.

7. The power consumption optimization control method according to claim 1, characterized in that, In step S5, the constraint projection correction is performed by minimizing the distance between the actual control quantity and the rolling optimization control quantity, and the correction control quantity is obtained within the feasible domain that satisfies the lower limit of terminal pressure difference, the minimum frequency of the circulating pump, the maximum frequency of the circulating pump, the maximum rate of change of frequency, and the upper limit of power distribution. When any key sensor data fails or the optimization result is not feasible, the system switches to the local backup control strategy based on terminal pressure difference.

8. A power consumption optimization control system for a heat exchange station power distribution system, characterized in that, The system includes a data acquisition module, a baseline prediction module, a heat flexibility mapping module, a sensitivity identification module, a rolling optimization module, a real-time correction module, a heat deficit management module, and a control execution module. The data acquisition module acquires power distribution and heating operation data. The baseline prediction module generates baseline heat load and baseline electrical load. The heat flexibility mapping module generates upward and downward power adjustment boundaries for power distribution based on heat balance, terminal pressure difference, and pump operation constraints. The sensitivity identification module establishes an electro-thermal-hydraulic sensitivity matrix between the control quantity and electrical power, heating power, and terminal pressure difference. The rolling optimization module forms a control sequence for pre-peak preheating, peak reduction, and post-peak replenishment within the power distribution boundaries. The real-time correction module corrects the control quantity based on real-time operating deviations and the electro-thermal-hydraulic sensitivity matrix. The heat deficit management module generates a heat deficit based on the difference between baseline heating power and actual heating power, and limits post-peak replenishment power based on remaining power distribution capacity and online electro-thermal conversion coefficients. The control execution module is used to send the corrected control quantity to the circulating pump frequency converter and / or the adjustable electrical equipment in the station.

9. The power consumption optimization control system according to claim 8, characterized in that, The data acquisition module includes an energy acquisition terminal installed on the main incoming line and adjustable load branch of the heat exchange station, as well as temperature, pressure and flow acquisition terminals installed on the secondary network supply side, return side and the most unfavorable loop. The control execution module is connected to the circulating pump frequency converter via an industrial communication bus or Ethernet, and retains the original PLC pressure closed loop as a backup control channel in case of optimized control failure.

10. The power consumption optimization control system according to claim 8, characterized in that, The rolling optimization module updates the equipment control sequence for multiple future control periods using a first time scale, and the real-time correction module performs constraint projection correction based on the real-time terminal pressure difference and real-time power using a second time scale smaller than the first time scale; the first time scale is 5 to 30 minutes, and the second time scale is 10 to 120 seconds.

Images