Water chilling unit operation control system and method based on artificial intelligence

Abstract

The invention relates to the field of intelligent control systems, and particularly discloses a water chilling unit operation control method based on artificial intelligence, which comprises the following steps: S1, data acquisition: respectively acquiring operation parameters of each water chilling unit, a chilled water pump, a cooling water pump and a cooling tower in an air conditioning system in real time by utilizing a plurality of sensors, outdoor meteorological parameters and environment parameters of all rooms are collected at the same time; s2, load prediction: establishing a load prediction model by using a neural network according to the collected historical outdoor meteorological parameters, the environment parameters of each room and the corresponding air conditioner load data, and predicting the air conditioner load of the building; s3, performance parameter determination: establishing a performance attenuation model of the water chilling unit, and correcting the COP to obtain real-time dynamic performance parameters of the unit; s4, control strategy generation: optimizing the energy consumption of the water system to obtain an optimal operation strategy of the water system; and S5, controlling execution. By the adoption of the technical scheme, the optimal energy efficiency can be achieved, accurate matching of supply and demand of the cooling capacity can be achieved, and user experience is improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control systems, and in particular to an artificial intelligence-based chiller operation control system and method. Background Art

[0002] In the air conditioning system of modern buildings, chillers, as the core refrigeration equipment, play a vital role. Typically, an air conditioning system consists of multiple chillers, chilled water pumps, cooling water pumps, cooling towers, control systems, and air conditioning terminals installed in each room. These devices work together to meet the cooling needs of different areas within the building. In large public buildings, air conditioning system energy consumption accounts for 40% to 60% of the total building energy consumption, and water systems (including chillers, chilled water pumps, cooling water pumps, and cooling towers) account for approximately 60% to 80% of the total air conditioning system energy consumption. Therefore, the energy-efficient operation of air conditioning water systems plays a key role in achieving building energy conservation and emission reduction.

[0003] However, in the actual operation of chillers, pursuing optimal energy efficiency and providing the best user experience present numerous challenges, making it difficult to achieve a balance between the two. From an energy efficiency perspective, traditional control strategies in the industry often aim for theoretically optimal energy efficiency. A key energy efficiency metric, the chiller's coefficient of performance (COP), serves as a crucial basis for control decisions. COP reflects the ratio of a chiller's cooling capacity to its input power. In theory, optimizing energy efficiency can be achieved by adjusting chiller operating parameters under specific operating conditions according to the factory-calibrated COP curve. However, the reality is much more complex. Chillers gradually degrade over time, with frequent problems such as heat exchanger fouling and reduced compressor efficiency. These factors can cause the unit's actual COP to fluctuate, deviating from the initial design value. Consequently, even if the control system continues to operate according to the established strategy based on the theoretical COP, the ultimate goal is only theoretically optimal energy efficiency, not the optimal efficiency that best reflects the chiller's current operating conditions, resulting in significant hidden energy waste. The air conditioning water system is a complex system. Each device works independently but is closely related to each other, and the relevant parameters are strongly coupled. Therefore, it is necessary to optimize and control the water system's chillers, chilled water pumps, cooling water pumps, and cooling towers to achieve the lowest total energy consumption for cooling the water system.

[0004] On the other hand, from the perspective of user experience, the functional and usage characteristics of each room in a building vary significantly, with frequent personnel turnover. This results in a dynamic change in the cooling load required for each room. The cooling demand varies significantly between rooms and at different times. For example, office areas are densely populated during daytime working hours, resulting in a high cooling load demand; while conference rooms experience intermittent peaks in cooling load depending on the meeting schedule. Without the introduction of advanced predictive methods, such as neural network technology, into the control system, relying solely on traditional static load estimation methods would make it impossible to accurately predict the cooling load demand for each room at the next moment. This can easily lead to a mismatch between cooling supply and actual demand, resulting in some rooms being overcooled or overheated, seriously affecting user comfort, wasting energy, and failing to achieve an optimal user experience.

[0005] Therefore, there is an urgent need for an artificial intelligence-based chiller operation control system and method that can achieve optimal energy efficiency of the air-conditioning water system, accurately match cooling supply and demand, and improve user experience. Summary of the Invention

[0006] The present invention provides an artificial intelligence-based chiller operation control system and method, which can not only achieve optimal energy efficiency of the air-conditioning water system and accurately match the supply and demand of cooling capacity, but also improve the user experience.

[0007] In order to solve the above technical problems, this application provides the following technical solutions:

[0008] The chiller operation control method based on artificial intelligence includes:

[0009] S1 Data Collection: Multiple sensors are used to collect real-time operating parameters of each chiller, chilled water pump, cooling water pump, and cooling tower in the air-conditioning system, as well as outdoor meteorological parameters and environmental parameters of each room;

[0010] S2 Load Forecasting: Based on the collected historical outdoor meteorological parameters, the environmental parameters of each room, and the corresponding air conditioning load data, a neural network is used to establish an air conditioning load forecasting model. After the model is calibrated, the collected outdoor meteorological parameters at the current moment, the outdoor meteorological parameter forecast values for a certain period of time in the future, the environmental parameters of each room, and the historical load data are input into the pre-trained neural network model. The neural network model is used to predict the cooling load required for each room in a preset time period in the future, and the total cooling load forecast required by the chiller system is obtained by summarizing the total cooling load required by the chiller system.

[0011] S3 Performance Parameter Determination: Based on the operating parameters, combined with the chiller's operating time, historical maintenance records, and the chiller's COP values at different load rates calculated and measured in real time, a performance degradation evaluation model for the chiller is established. The current performance degradation coefficient of the chiller is calculated using the performance degradation evaluation model, and the initial performance coefficient curve of the chiller is corrected based on the performance degradation coefficient to determine the current actual dynamic performance coefficient COP value.

[0012] S4 Control Strategy Generation: Calculate the energy consumption of the chiller based on the total cooling load forecast, the chiller start / stop combination, the load rate of each chiller, and the current actual dynamic coefficient of performance (COP) value, establish energy consumption models for the chilled water pump, cooling water pump, and cooling tower, and use an optimization algorithm to generate the optimal operation control strategy for each device in the air conditioning water system;

[0013] S5 control execution: transmitting the optimal operation control strategy to the controller corresponding to the air-conditioning system through the communication module to control the chiller, the chilled water pump, the cooling water pump and the cooling tower fan to operate according to the optimal operation control strategy.

[0014] Furthermore, the operating parameters include temperature, pressure, flow and frequency parameters; the outdoor meteorological parameters include outdoor dry-bulb temperature, relative humidity, solar radiation intensity, and wind speed parameters; the indoor environmental parameters include room temperature, relative humidity, number of people, indoor lighting, and electrical equipment parameters; the optimal operation control strategy includes the start-stop combination of each chiller, the load factor of each chiller, the set values of the chilled water pump speed, the cooling water pump speed, and the cooling tower fan speed, as well as the chilled water supply temperature and the cooling water inlet temperature set value adjustment instructions;

[0015] In S2, the collected environmental parameters X of each room are i,t , historical load data Q i,t；k and outdoor meteorological parameters Y t , where i represents the room number, i = 1, 2, ..., n, n is the total number of rooms; t represents the current time; k = 1, 2, ..., m, m is the number of selected historical time steps, and then feature extraction and preprocessing are performed;

[0016] Environmental parameters of each room X i,t It can be expressed as a multidimensional vector, namely in, Represent different environmental parameters of room i at time t; outdoor meteorological parameter Y t It can also be expressed as a multidimensional vector, that is, in, Represent different outdoor meteorological parameters at time t; historical load data Q i,t；kis the cooling load value of room i in the past m time steps; outdoor meteorological parameters, indoor environmental parameters and historical load data are preprocessed, including normalization, and the data are mapped to the [0,1] interval.

[0017] Furthermore, the normalization process is as follows:

[0018] For indoor environmental parameters Its normalized value for:

[0019]

[0020] Among them, min(x j ) and max(x j ) are the minimum and maximum values of the environmental parameters in the historical data respectively.

[0021] For outdoor meteorological parameters Its normalized value for:

[0022]

[0023] in, and are the minimum and maximum values of the environmental parameter in the historical data respectively.

[0024] Furthermore, for the historical load data Q i,t；k , its normalized value for:

[0025]

[0026] Among them, min(Q) and max(Q) are the minimum and maximum values of the historical cooling load data of all rooms, respectively.

[0027] Furthermore, the pre-processed indoor environment parameters Outdoor meteorological parameters and historical load data Combined into input vector I i,t ,Right now:

[0028]

[0029] Then the input vector I i,t Input into the pre-trained long short-term memory network, i.e. LSTM model.

[0030] Furthermore, the LSTM model consists of an input layer, multiple LSTM hidden layers, and an output layer;

[0031] In the LSTM hidden layer, the calculation process of each LSTM unit is as follows:

[0032] Input gate i t :

[0033] i t =σ(W ii I i,t +W hi h t；1 +b i )

[0034] Forget Gate f t :

[0035] f t =σ(W if I i,t +W hf h t；1 +b f )

[0036] Cell status update

[0037]

[0038] Cell State C t :

[0039]

[0040] Output gate o t :o t =σ(W io I i,t +W ho h t；1 +b o )

[0041] Hidden state h t :

[0042] h t =o t ⊙tanh(C t )

[0043] Among them, σ is the sigmoid function, tanh is the hyperbolic tangent function W ii 、W hi 、W if 、W hf 、W ic 、W hc 、W io 、W ho is the weight matrix, b i 、b f 、b c 、bo is the bias vector, ⊙ represents element-by-element multiplication, h t；1 is the hidden state of the previous moment, C t；1 is the cell state at the previous moment;

[0044] After calculations of multiple LSTM hidden layers, the output h of the last hidden layer is t Input to the output layer, and obtain the cooling load forecast value required by room i in the future preset time period Δt through linear transformation

[0045]

[0046] Among them, W out is the output layer weight matrix, b out is the output layer bias vector;

[0047] Finally, the cooling load forecast values of all rooms are summarized to obtain the total cooling load forecast value required by the chiller system.

[0048] Furthermore, in S3, first, key features are extracted from the collected operating parameters of each component of the chiller. Assume that the chiller has m key operating parameters, and the value of the jth operating parameter at time t is recorded as x j,t , where j = 1, 2, ..., m; at the same time, record the cumulative running time T of the chiller from the time it was put into use to time t t , and the maintenance impact factor S quantified based on historical maintenance records t ; Use the multiple linear regression model to calculate the current performance attenuation coefficient α of the chiller t , considering the comprehensive impact of operating parameters, operating time and maintenance factors on unit performance degradation, its expression is:

[0049]

[0050] Among them, β0 is a constant term, β j is the regression coefficient, j=1,2,…,m+2; the regression coefficient can be obtained by training historical data;∈ t is a random error term, which obeys a normal distribution with a mean of 0;

[0051] Use the least squares method to estimate the regression coefficient. Assume there are N sets of historical data. The goal is to minimize the sum of squared errors S:

[0052]

[0053] By calculating S with respect to β0, β1, ..., β m:2By finding the partial derivative and setting it to 0, we can get a system of linear equations. Solving this system of equations can give us the estimated values of the regression coefficients.

[0054] The initial performance coefficient curve of the chiller is represented by a polynomial function. Let the initial performance coefficient COP0 be the partial load rate PLR of the chiller, the chilled water supply temperature T chws , cooling water inlet temperature T cws The function is expressed as:

[0055] COP0(PLR,T chws ,T cws )

[0056] =a0+a1PLR+a2(PLR) 2 +a3T chws +a4T 2 chws +a5T cws

[0057] +a6T 2 cws +a7PLR·T chws +a8PLR·T cws +a9T chws ·T cws +a 10 PLR·T cHws ·T cws

[0058] Among them, a i are polynomial coefficients, i = 0, 1, ..., 10, obtained by fitting the factory test data or experimental data of the unit;

[0059] Based on the calculated performance attenuation coefficient α t· , correct the initial performance coefficient curve to obtain the actual performance coefficient curve COP at the current moment t (PLR,T chws ,T cws ), the correction formula is:

[0060] COP t (PLR,T chws ,T cws )=α t ·COP0(PLR,T chws ,T cws )

[0061] Performance attenuation coefficient α t As a correction factor, the initial performance coefficient curve is corrected;

[0062] To get the actual performance coefficient curve COP at the current moment t (PLR t ,T chws,t ,T cws,t ), according to the actual operating conditions of the chiller at the current moment, determine the key operating parameter value PLR,T corresponding to COP chws ,T cws , and substituting it into the corrected performance coefficient curve equation, we can get the current actual dynamic performance coefficient COP actual,t :COP actual,t =COP t (PLR t ,T cHws,t ,T cws,t )=α t COP0(PLR t ,T chws,t ,T cws,t ).

[0063] Furthermore, in said S4, the total energy consumption P of the air conditioning water system is minimized. total As the goal, while considering the total cooling load forecast value Demand; total energy consumption P total The energy consumption P of each chiller chiller , Energy consumption of chilled water pump P chwp , Energy consumption of cooling water pump P cwp , cooling tower fan energy consumption P tfan composition.

[0064] Furthermore, the total energy consumption of each chiller is P chiller It can be expressed as:

[0065] COP i,0 =a0+a1PLR i +a2(PLR i ) 2 +a3T chws +a4T 2 chws +a5T cws

[0066] +a6T 2 cws +a7PLR i ·T cHws +a8PLR i ·T cws +a9T cHws ·T cws +a 10 PLR i ·T chws ·T cws

[0067]

[0068] Among them, N chiler is the total number of chillers, P chiller,i is the power of the i-th chiller, Q chiller,i is the cooling capacity of the i-th chiller, COP i,0 is the COP value of the i-th chiller at the current moment obtained according to the initial performance curve, α t is the performance attenuation coefficient; the energy consumption of the refrigeration pump P chwp and cooling water pump energy consumption P cwp and the speed of the water pump n pump Considering the operating efficiency of the water pump, according to the similarity law of the water pump, the energy consumption of the chilled water pump and the cooling water pump can be approximately expressed as:

[0069] P chwp =b1·ω chwp ·m chw 2 +b2·ω chwp 2 ·m chw +b3·ω chwp 3

[0070] P cwp =c1·ω cwp ·m cw 2 +c2·ω cwp 2 ·m cw +c3·ω cwp 3

[0071]

[0072] Among them, P chwp and P cwp are the power of the chilled water pump and the cooling water pump respectively, n chwp and n cwp is the speed of the chilled water pump and cooling water pump when they are running, n chwp,nom and n cwp,nom are the speeds of the chilled water pump and cooling water pump at rated operating conditions, m chw and m cw are the chilled water pump flow rate and the cooling water pump flow rate, ω chwp and ω cwp are the speed ratios of the chilled water pump and the cooling water pump, respectively.

[0073] Energy consumption of cooling tower P tfan The fan speed n of the cooling towertfan The energy consumption expression can be obtained based on the empirical formula and experimental data fitting:

[0074] P tfan =P tfan,nom (e0+e1PLR tfan +e2(PLR tfan ) 2 +e3(PLR tfan ) 3 )

[0075]

[0076] Among them, P tfan is the actual power of the cooling tower, P tfan,nom is the rated power of the cooling tower, m ta is the actual air volume of the cooling tower, m ta，nom is the rated air volume of the cooling tower, e i is the model parameter, i = 0, 1, 2, 3, which can be obtained through the performance test of the cooling tower and data regression analysis;

[0077] Then the objective function is:

[0078]

[0079] Then set constraint 1, cooling load constraint: the total cooling capacity Q provided by each chiller chiller The total cooling load forecast value must be met Right now:

[0080]

[0081] Among them, Q chiller,i is the cooling capacity of the i-th chiller in the current operating state;

[0082] Set constraint 2, equipment operation range constraint: the start and stop status u of each chiller i ∈{0,1}; the speed of the chilled water pump n chwp The minimum speed n allowed chwp,min and maximum speed n chwp,max Between, that is, n chwp,min ≤n chwp ≤n chwp,max ; Cooling water pump speed n cwp Also, it must be at its minimum speed n cwp,min and maximum speed n cwp,max Between, that is, n cwp,min ≤n cwp≤ n cwp,max ; Cooling tower fan speed n tfan The minimum speed n must be mettfan,min and maximum speed n tfan,max The limit, that is, n tfan,min ≤n tfan≤ n tfan,max ;

[0083] Set constraint 3, the chilled water supply temperature and cooling water inlet temperature range constraints of the chiller: chilled water supply temperature T chws and cooling water inlet temperature T cws Between their lowest and highest temperatures, respectively,

[0084] T chws,min ≤T chws ≤T chws,max

[0085] T cws,min ≤T cws ≤T cws,max

[0086] Set constraint 4, the flow range constraint of the chilled water pump and cooling water pump: the flow of each chilled water pump and cooling water pump is between its minimum flow and maximum flow, that is,

[0087] m chw,j,min ≤m chw,j ≤m chw,j,max

[0088] m chw,j,min ≤m chw,j ≤m chw,j,max

[0089] Set constraint 5, mutual constraints between devices:

[0090]

[0091] Among them, T chws and T chwr are the supply and return temperatures of chilled water, T cws and T cwr are the temperatures of cooling water entering and leaving the chiller, C pw is the specific heat of water.

[0092] The genetic algorithm is used to solve the above objective function under the constraints:

[0093] Initialize the population: Randomly generate a set of individuals that include the start and stop combinations of each chiller, the load rate of each chiller, the chilled water supply temperature, the cooling water inlet temperature, the chilled water pump speed, the cooling water pump speed, and the cooling tower fan speed to form the initial population;

[0094] Fitness calculation: For each individual in the population, its fitness value is calculated according to the objective function. The smaller the fitness value, the better the individual.

[0095] Selection operation: Roulette wheel selection method is used to select a certain number of individuals to enter the next generation according to their fitness values;

[0096] Crossover operation: Perform crossover operation on the selected individuals to generate new individuals;

[0097] Mutation operation: perform mutation operations on newly generated individuals to increase the diversity of the population;

[0098] Termination condition judgment: When the preset termination condition of reaching the maximum number of iterations or the fitness value convergence is met, the iteration is stopped and the optimal individual is output. The start and stop combination of each chiller, the load rate of each chiller, the chilled water supply temperature, the cooling water inlet temperature, the chilled water pump speed, the cooling water pump speed, and the cooling tower fan speed corresponding to the optimal individual are the optimal operation control strategy.

[0099] The basic solution's principles and benefits are as follows: During chiller operation, the system uses multiple sensors to collect real-time operating parameters from key components (including multiple chillers, chilled water pumps, cooling water pumps, and cooling towers). It also collects outdoor meteorological parameters and room-by-room environmental parameters. This rich data forms the foundation for subsequent analysis and decision-making, comprehensively reflecting the operating status of the air conditioning water system and user needs.

[0100] A pre-trained neural network model takes as input collected outdoor meteorological parameters, room environmental parameters, and historical load data. Neural networks possess powerful nonlinear mapping capabilities, enabling them to learn the complex relationships between environmental parameters and cooling load. By studying and analyzing historical data, the cooling load required for each room within a preset future time period is predicted. The predicted values for each room are then aggregated to produce a predicted total cooling load for the chiller system. Load forecasting also incorporates the air conditioning system's operating day characteristics, enabling classification of weekdays and holidays to improve load forecast accuracy. This step provides an accurate basis for load demand when subsequently determining the chiller's operating strategy.

[0101] Considering that chillers experience performance degradation during long-term operation, their coefficient of performance (COP) will change. This method combines collected operating parameters of each device, unit operating time, and historical maintenance records with an established chiller performance degradation assessment model to calculate the current COP. This coefficient reflects the extent to which unit performance degradation affects performance. The initial COP curve of the chiller is then corrected based on this coefficient to obtain the current actual dynamic COP value. This more accurately reflects the unit's energy efficiency in its current state.

[0102] The energy consumption of the chillers is calculated based on the total cooling load forecast and dynamic COP value. Energy consumption models for the chilled water pump, cooling water pump, and cooling tower are also established. Optimization algorithms (such as genetic algorithms) are then used to find the optimal operation control strategy for each device in the air conditioning water system. The optimization algorithm solves the objective function (e.g., minimizing total energy consumption) under a series of constraints (e.g., cooling load requirements, equipment operating range restrictions, etc.) to determine the start / stop combinations for each chiller, the load factor for each chiller, the chilled water supply temperature, the cooling water inlet temperature, the chilled water pump speed, the cooling water pump speed, and the cooling tower fan speed setpoints. This ensures that the air conditioning water system achieves optimal system energy efficiency while meeting user cooling load requirements.

[0103] The generated optimal operation control strategy is transmitted to the controller corresponding to the air-conditioning system through the communication module. The controller controls the chiller, chilled water pump, cooling water pump and cooling tower fan according to the received instructions to operate according to the optimal strategy, thereby realizing real-time and precise control of the air-conditioning water system.

[0104] Through accurate load forecasting and dynamic performance coefficient calculation, combined with the optimal operation control strategy generated by the optimization algorithm, the air conditioning water system can operate precisely according to actual demand, avoiding energy waste caused by overcooling or inefficient operation under traditional control methods. Due to the complex and nonlinear characteristics of the air conditioning water system, the lowest energy consumption of a single device does not necessarily achieve the lowest energy consumption of the water system. By optimizing the water system globally, global optimal control of the air conditioning water system is achieved. Accurate load forecasting enables the chiller to dynamically adjust the cooling capacity according to the actual needs of each room, avoiding overcooling or overheating in some rooms, providing users with a more comfortable and stable indoor environment and improving the user experience. While meeting user cooling load requirements, it effectively reduces the total energy consumption of the air conditioning water system and improves energy utilization efficiency, saving businesses and society significant energy costs and promoting energy conservation and emission reduction in buildings.

[0105] Taking into account the performance degradation of chillers, a performance degradation assessment model is used to monitor performance changes in real time and adjust operating strategies accordingly, achieving real-time dynamic optimization of air conditioning water system performance. This prevents excessive or inappropriate operation of equipment in a degraded state, reduces equipment wear and failure, thereby extending the life of the chiller and reducing maintenance costs and replacement frequency.

[0106] The neural network model can learn the complex relationships between outdoor meteorological parameters, indoor environmental parameters, and cooling load, adapting to varying environmental conditions and changing user needs. Furthermore, the calculation of the dynamic coefficient of performance (COP) enables the system to assess the performance degradation and performance changes of the chiller in real time, accurately predicting the chiller's real-time COP and energy consumption. Energy consumption models for the chilled water pump, cooling water pump, and cooling tower are also established, enabling global optimization of the air conditioning water system. This ensures stable and efficient operation under varying operating conditions, enhancing the system's adaptability and robustness.

[0107] In summary, the present invention achieves the goal of optimizing the energy efficiency of the air conditioning water system, achieving accurate matching of supply and demand, and improving user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0108] Figure 1 The flowchart of the chiller operation control method based on artificial intelligence. DETAILED DESCRIPTION

[0109] The following is further described in detail through specific implementation methods:

[0110] The embodiment is basically as follows Figure 1 As shown, the chiller operation control method based on artificial intelligence includes:

[0111] S1 Data Collection: Multiple sensors are used to collect real-time operating parameters of each chiller, chilled water pump, cooling water pump, cooling tower fan, chilled water flow, cooling water flow, chilled water supply and return water temperature, and cooling water inlet and outlet temperature of the air-conditioning system. At the same time, outdoor meteorological parameters and environmental parameters of each room are collected. The operating parameters include temperature, pressure, flow and frequency. The outdoor meteorological parameters include outdoor temperature, relative humidity, solar radiation intensity, wind speed, etc. The environmental parameters include room temperature, relative humidity, number of people, equipment and lighting operating parameters. A commercial complex is used as an example for detailed description. The commercial complex has 30 floors above ground and an air-conditioned area of 45,000 m 2 The air conditioning system has a designed cooling load of 5,700 kW and uses three centrifugal chillers, four chilled water pumps (three in service and one backup), four cooling water pumps (three in service and one backup), and three cooling towers. Each chiller has a rated cooling capacity of 1,950 kW and a rated power of 336 kW.

[0112] S2 load prediction: Based on the collected historical outdoor meteorological parameters, environmental parameters of each room and corresponding air-conditioning load data, an air-conditioning load prediction model is established using a neural network; after the model is calibrated, the collected outdoor meteorological parameters at the current moment, the forecast values of outdoor meteorological parameters for a certain time step in the future, the environmental parameters of each room and the historical load data are input into a pre-trained neural network model, and the cooling load required for each room in a preset time period in the future is predicted by the neural network model, and the total cooling load prediction value required to be provided by the chiller system is obtained by summarizing; and jumping to S3 or S4 according to the preset conditions, the preset conditions are whether the total running time interval of the unit from the last execution of S3 to the current moment meets the preset standard (240h in this embodiment), if the interval is greater than 240h, execute S3; if the interval is less than 240h, jump to S4 (when executing S4, the parameters related to S3 are selected as the parameters when S3 was last executed).

[0113] S3 Performance Parameter Determination: Based on the collected operating parameters of each device in the air conditioning system, combined with the chiller's operating time, historical maintenance records, and the chiller's COP values at different load rates obtained through real-time measurement, a performance degradation evaluation model for the chiller is established. The current performance degradation coefficient of the chiller is calculated using the performance degradation evaluation model, and the initial performance coefficient curve of the chiller is corrected based on the performance degradation coefficient to determine the current actual dynamic performance coefficient COP value.

[0114] S4 control strategy generation: Calculate the energy consumption of the chiller based on the total cooling load forecast value, the start / stop combination of the chiller, the load rate of each chiller, and the determined current actual dynamic coefficient of performance (COP) value, establish energy consumption models for the chilled water pump, cooling water pump, and cooling tower, and use an optimization algorithm to generate the optimal operation control strategy for each device in the air-conditioning water system. The optimal operation control strategy includes the start / stop combination of each chiller, the load rate of each chiller, the chilled water pump speed, the cooling water pump speed, and the cooling tower fan speed set value, as well as the chilled water supply temperature and the cooling water inlet temperature set value and other adjustment instructions;

[0115] S5 control execution: transmitting the optimal operation control strategy to the controller corresponding to the chiller through the communication module to control the chiller, the chilled water pump, the cooling water pump and the cooling tower fan to operate according to the optimal operation control strategy.

[0116] Specifically, in S2, the collected environmental parameters X of each room are i,t , historical load data Q i,t；k and outdoor meteorological parameters Y t, where i represents the room number, i = 1, 2, ..., n, n is the total number of rooms; t represents the current time; k = 1, 2, ..., m, m is the number of selected historical time steps, and then feature extraction and preprocessing are performed;

[0117] Environmental parameters of each room X i,t It can be expressed as a multidimensional vector, namely in, Represent different environmental parameters of room i at time t; outdoor meteorological parameter Y t It can also be expressed as a multidimensional vector, that is, in, Represent different outdoor meteorological parameters at time t; historical load data Q i,t；k is the cooling load value for room i over the past m time steps. Preprocess the outdoor meteorological parameters, indoor environmental parameters, and historical load data, including normalization, to map the data to the [0, 1] interval to improve the training efficiency and stability of the neural network. The past year's data is selected as historical data, with a time step of 15 minutes.

[0118] For room 10, the temperature parameters Obtain the minimum value of the temperature parameter through historical data statistics Maximum If the temperature of the room at the current moment t is 25℃.

[0119] The normalization process is as follows:

[0120] For indoor environmental parameters Its normalized value for:

[0121]

[0122] Among them, min(x j ) and max(x j ) are the minimum and maximum values of the environmental parameters in the historical data respectively.

[0123] After entering the value,

[0124] For outdoor meteorological parameters Its normalized value for:

[0125]

[0126] in, and are the minimum and maximum values of the environmental parameter in the historical data respectively.

[0127] Take the outdoor temperature as an example. For example, the outdoor temperature is 15℃ at the lowest and 35℃ at the highest. The current outdoor temperature is 25℃. After entering the value,

[0128] For historical load data Q i,t；k , its normalized value for:

[0129]

[0130] Among them, min(Q) and max(Q) are the minimum and maximum values of the historical cooling load data of all rooms, respectively.

[0131] For the historical load data, assume that the minimum value min(Q) in the historical cooling load data for this room is 2.88kW and the maximum value max(Q) is 9.09kW. The cooling load of room 10 at a certain point in the past was 6.65kW. The normalized value is:

[0132]

[0133] The pre-processed indoor environmental parameters Outdoor meteorological parameter Y t and historical load data Combined into input vector I i,t ,Right now:

[0134] Then the input vector I i,t Input into the pre-trained long short-term memory network, i.e. LSTM model.

[0135] The LSTM model consists of an input layer, multiple LSTM hidden layers, and an output layer;

[0136] In the LSTM hidden layer, the calculation process of each LSTM unit is as follows:

[0137] Input gate i t :

[0138] i t =σ(W ii I i,t +W hi h t；1 +b i )

[0139] Forget Gate f t :

[0140] f t =σ(W if I i,t +W hfh t；1 +b f )

[0141] Cell status update

[0142]

[0143] Cell State C t :

[0144]

[0145] Output gate o t :o t =σ(W io I i,t +W ho h t；1 +b o )

[0146] Hidden state h t :

[0147] h t =o t ⊙tanh(C t )

[0148] Among them, σ is the sigmoid function, tanh is the hyperbolic tangent function, W ii 、W hi 、W if 、W hf 、W ic 、W hc 、W io 、W ho is the weight matrix, b i 、b f 、b c 、b o is the bias vector, ⊙ represents element-by-element multiplication, h t；1 is the hidden state of the previous moment, C t；1 is the cell state at the previous moment; after calculation of multiple LSTM hidden layers, the output h of the last hidden layer is t Input to the output layer, and obtain the cooling load forecast value required by room i in the future preset time period Δt (assuming it is 15 minutes) through linear transformation

[0149]

[0150] Among them, W out is the output layer weight matrix, b out is the output layer bias vector;

[0151] Finally, the cooling load forecast values of all rooms are summed up to get the total cooling load forecast value required by the chiller system

[0152]

[0153] Assume that after calculation, the total cooling load forecast value required by the chiller system in the next 15 minutes is 3600kW.

[0154] In S3, first, key features are extracted from the collected operating parameters of each component of the chiller. Assume that the chiller has m key operating parameters, and the value of the jth operating parameter at time t is recorded as x j,t , where j = 1, 2, ..., m; at the same time, record the cumulative running time T of the chiller from the time it was put into use to time t t = 500 hours, and the maintenance impact factor S quantified based on historical maintenance records t =0.9 (indicating that recent maintenance has a certain positive impact on unit performance); a multiple linear regression model is used to calculate the current performance attenuation coefficient α of the chiller t , considering the comprehensive impact of operating parameters, operating time and maintenance factors on unit performance degradation, its expression is:

[0155]

[0156] Among them, β0 is a constant term, β j is the regression coefficient, j=1,2,…,m+2; the regression coefficient can be obtained by training historical data;∈ t is a random error term, which obeys a normal distribution with a mean of 0;

[0157] Use the least squares method to estimate the regression coefficient. Assume there are 100 sets of historical data. The goal is to minimize the sum of squared errors S:

[0158]

[0159] By calculating S with respect to β0, β1, ..., β m:2 By taking the partial derivative and setting it to 0, we can get a system of linear equations. Solving this system of equations can give us the estimated values of the regression coefficients.

[0160] According to the actual operating conditions of the chiller at the current moment, determine the corresponding key operating parameter values PLR, T chws and T cws , after calculation, the performance attenuation coefficient α of the chiller at this moment is obtained t =0.95.

[0161] The initial performance coefficient curve of the chiller is represented by a polynomial function. Let the initial performance coefficient COP0 be the partial load rate PLR of the chiller, the chilled water supply temperature T chws , cooling water inlet temperature T cws The function is expressed as:

[0162] COP0(PLR,T chws ,T cws )

[0163] =a0+a1PLR+a2(PLR) 2 +a3T chws +a4T 2 chws +a5T cws +a6T 2 cws +a7PLR·T chws +a8PLR·T cws +a9T chws ·T cws +a 10 PLR·T chws ·T cws

[0164] Among them, a1 to a 10 are polynomial coefficients, obtained by fitting the factory test data or experimental data of the unit;

[0165] In this embodiment, the following are selected:

[0166]

[0167] Correct the initial performance coefficient curve to obtain the actual performance coefficient COP at the current moment i,t (PLR,T chws ,T cws ), the correction formula is:

[0168] COP i,t (PLR,T chws ,T cws )=α t COP i,0 (PLR,T chws ,T cws )

[0169] The performance attenuation coefficient α of the current chiller t Substituting into the corrected performance coefficient curve equation, the actual dynamic performance coefficient COP can be obtained. actual,t :

[0170] COP actual,i,t =αt COP i,0 (PLR t ,T chws,t ,T cws,t )=0.95*COP i,0 (PLR t ,T chws,t ,T cws,t )

[0171] The current load rate of chiller 2 is PLR2 = 0.9, T chws,t =7℃, T cws,t =32℃, then

[0172]

[0173] In S4, the total energy consumption of the air conditioning system is minimized total As the goal, while considering the total cooling load forecast value Demand; total energy consumption P total The energy consumption P of each chiller chiller , Energy consumption of chilled water pump P chwp , Energy consumption of cooling water pump P cwp , cooling tower fan energy consumption P tfan composition.

[0174] Energy consumption of each chiller P chiller It can be expressed as:

[0175]

[0176]

[0177] Among them, N chiler is the total number of chillers, P chiller,i is the power of the i-th chiller, COP i,0 is the COP value of the i-th chiller at the current moment obtained according to the initial performance curve, α t is the performance attenuation coefficient.

[0178] Energy consumption of chilled water pump P chwp and cooling water pump energy consumption P cwp and the speed of the water pump n pump Considering the operating efficiency of the water pump, according to the similarity law of the water pump, the energy consumption of the refrigeration pump and the cooling water pump can be approximately expressed as:

[0179] P chwp =b1·ω chwp ·m chw 2 +b2·ω chwp 2 ·mchw +b3·ω chwp 3

[0180] P cwp =c1·ω cwp ·m cw 2 +c2·ω cwp 2 ·m cw +c3·ω cwp 3

[0181]

[0182] Among them, P chwp,nom and P cwp,nom are the power of chilled water pump and cooling water pump under rated working conditions, n chwp and n cwp is the speed of the chilled water pump and cooling water pump when they are running, n chwp,nom and n cwp,nom are the speeds of the chilled water pump and cooling water pump at rated operating conditions, m chw and m cw are the chilled water flow and cooling water flow respectively, ω chwp and ω cwp are the speed ratios of the chilled water pump and the cooling water pump respectively;

[0183] In this case, the experimental data of the chilled water pump and the cooling water pump are fitted to obtain

[0184] P chwp =-0.0001·ω chwp ·m chw 2 +0.0551·ω chwp 2 ·m chw +21.3693·ω chwp 3

[0185] P cwp =-0.0001·ω cwp ·m cw 2 +0.0618·ω cwp 2 ·m cw +23.9591·ω cwp 3

[0186] Energy consumption of cooling tower P tfan The fan speed of the cooling tower is n tfanThe energy consumption expression can be obtained based on the empirical formula and experimental data fitting:

[0187] P tfan =P tfan,nom (e0+e1PLR tfan +e2(PLR tfan ) 2 +e3(PLR tfan ) 3 )

[0188]

[0189] Among them, P tfan is the actual power of the cooling tower, P tfan,nom is the rated power of the cooling tower, m a is the actual air volume of the cooling tower, m a,nom is the rated air volume of the cooling tower, e i are model parameters, i = 0, 1, 2, 3, which can be obtained through performance testing of the cooling tower and data regression analysis. In this embodiment, the power of a single cooling tower fan is 15kW, e0 = 66.532, e1 = -282.59, e2 = 377.7, e3 = -146.51, that is,

[0190] P tfan =15*(66.532+282.59PLR tfan +377.7(PLR tfan ) 2 +146.51(PLR tfan ) 3 )

[0191] Then the objective function is:

[0192]

[0193] Then set constraint 1, cooling load constraint: the total cooling capacity Q provided by each chiller chiller The total cooling load forecast value must be met (3600kW in this embodiment), that is:

[0194]

[0195] Right now

[0196] Its, Q chiller,i is the cooling capacity of the i-th chiller in the current operating state;

[0197] Set constraint 2, equipment operation range constraint: the start and stop status u of each chiller i∈{0,1}; the speed of the chilled water pump n chwp The minimum speed n allowed chwp,min =725r / min and maximum speed n chwp,max =1450r / min, i.e. 725r / min≤n chwp ≤1450r / min; cooling tower fan speed n tfan The minimum speed n must be met tfan,min =300r / min and maximum speed n cfan,max =600r / min limit, that is, 300r / min≤n tfan ≤600r / min, cooling water pump speed n cwp Also, it must be at its minimum speed n cwp,min =725r / min and maximum speed n cwp,max =1450r / min, i.e. 725r / min≤n cwp ≤1450r / min;

[0198] Set constraint 3, the chilled water supply temperature and cooling water inlet temperature range constraints of the chiller: the chilled water supply temperature range is 5~12℃, that is, 7℃≤T chws ≤12℃; cooling water inlet temperature T cws The range is 19~33℃, that is, 19℃≤T cws ≤33℃;

[0199] On the basis of satisfying constraints 4 and 5, the genetic algorithm is used to solve the above objective function under the constraints:

[0200] Initialize the population: Randomly generate 100 individuals containing the start-stop combinations of each chiller, the load rate of each chiller, the speed of the chilled water pump, cooling water pump and cooling tower fan, and the set values of the chilled water supply temperature and cooling water inlet temperature to form the initial population.

[0201] Fitness calculation: For each individual in the population, its fitness value is calculated according to the objective function. The smaller the fitness value, the better the individual.

[0202] Selection operation: Use the roulette wheel selection method to select 50 individuals to enter the next generation based on their fitness values.

[0203] Crossover operation: Perform a crossover operation on the selected 50 individuals to generate 50 new individuals.

[0204] Mutation operation: Perform mutation operation on the newly generated 50 individuals to increase the diversity of the population.

[0205] Termination condition judgment: Set the maximum number of iterations to 200. When the 150th iteration is reached, the fitness value converges, the iteration is stopped, and the optimal individual is output.

[0206] In this embodiment, the control strategy corresponding to the optimal individual is: start chiller 2 and chiller 3 (u1 = 0, u2 = 1, u3 = 1), the load rate PLR of the two chillers is 0.923, and the speed of the corresponding two chilled water pumps n chwp =1360r / min, cooling water pump speed n cwp =1385r / min, cooling tower fan speed n tfan =535r / min, chilled water supply temperature T chws is 7.3℃, cooling water inlet temperature T cws It is 30.5℃.

[0207] The generated optimal operation control strategy is transmitted to the corresponding controller of the air conditioning system through a wireless communication module (such as Wi-Fi). After receiving the command, the controller controls the operation of chiller 2 and chiller 3. The load rate PLR of the two chillers is 0.923. The speed of the chilled water pump is adjusted to 1360r / min, the speed of the cooling water pump is 1385r / min, the speed of the cooling tower fan is set to 535r / min, and the chilled water supply temperature T is set to 0. chws Set to 7.3℃, cooling water inlet temperature T cws Set to 30.5℃.

[0208] Through the above steps, this artificial intelligence-based chiller operation control method can achieve efficient and intelligent control of the chiller and water system of the commercial office building, while meeting the cooling needs of the building, effectively reducing energy consumption and improving energy utilization efficiency.

[0209] The above are only embodiments of the present invention. The invention is not limited to the fields involved in this implementation case. Common knowledge such as the known specific structures and characteristics in the scheme is not described in detail here. Ordinary technicians in the relevant field are aware of all common technical knowledge in the technical field to which the invention belongs before the application date or priority date, can obtain all existing technologies in the field, and have the ability to apply conventional experimental means before that date. Ordinary technicians in the relevant field can improve and implement this scheme in combination with their own abilities under the inspiration given by this application. Some typical known structures or known methods should not become obstacles for ordinary technicians in the relevant field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several variations and improvements can be made, which should also be regarded as the scope of protection of the present invention. These will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. The chiller operation control method based on artificial intelligence is characterized by: include: S1 Data Collection: Multiple sensors are used to collect real-time operating parameters of each chiller, chilled water pump, cooling water pump, and cooling tower in the air-conditioning system, as well as outdoor meteorological parameters and environmental parameters of each room. S2 Load Forecasting: Based on the collected historical outdoor meteorological parameters, environmental parameters of each room and corresponding air conditioning load data, a neural network is used to establish an air conditioning load forecasting model; After the model is calibrated, the collected outdoor meteorological parameters at the current moment, the forecast values of outdoor meteorological parameters for a certain period of time in the future, the environmental parameters of each room, and the historical load data are input into the pre-trained neural network model. The neural network model is used to predict the cooling load required for each room in a preset time period in the future, and the total cooling load forecast required by the chiller system is summarized; S3 Performance Parameter Determination: Based on the operating parameters, combined with the chiller's operating time, historical maintenance records, and the chiller's COP values at different load rates calculated and measured in real time, a performance degradation evaluation model for the chiller is established. The current performance degradation coefficient of the chiller is calculated using the performance degradation evaluation model, and the initial performance coefficient curve of the chiller is corrected based on the performance degradation coefficient to determine the current actual dynamic performance coefficient COP value. S4 Control Strategy Generation: Calculate the energy consumption of the chiller based on the total cooling load forecast, the chiller start / stop combination, the load rate of each chiller, and the current actual dynamic coefficient of performance (COP) value, establish energy consumption models for the chilled water pump, cooling water pump, and cooling tower, and use an optimization algorithm to generate the optimal operation control strategy for each device in the air conditioning water system; S5 control execution: transmitting the optimal operation control strategy to the controller corresponding to the air-conditioning system through the communication module to control the chiller, the chilled water pump, the cooling water pump and the cooling tower fan to operate according to the optimal operation control strategy.

2. The chiller operation control method based on artificial intelligence according to claim 1 is characterized in that: The operating parameters include temperature, pressure, flow and frequency parameters; the outdoor meteorological parameters include outdoor dry-bulb temperature, relative humidity, solar radiation intensity, and wind speed parameters; the indoor environmental parameters include room temperature, relative humidity, number of people, indoor lighting, and electrical equipment parameters; the optimal operation control strategy includes the start-stop combination of each chiller, the load rate of each chiller, the set values of the chilled water pump speed, the cooling water pump speed, and the cooling tower fan speed, as well as the chilled water supply temperature and the cooling water inlet temperature set value adjustment instructions; In S2, the collected environmental parameters X of each room are i,t , historical load data Q i,t；k and outdoor meteorological parameters Y t , where i represents the room number, i = 1, 2, ..., n, n is the total number of rooms, t represents the current time; k = 1, 2, ..., m, m is the number of selected historical time steps, and then feature extraction and preprocessing are performed; Environmental parameters of each room X i,t It can be expressed as a multidimensional vector, namely in, Represent different environmental parameters of room i at time t; outdoor meteorological parameter Y t It can also be expressed as a multidimensional vector, that is, in, Represent different outdoor meteorological parameters at time t; historical load data Q i,t；k is the cooling load value of room i in the past m time steps; outdoor meteorological parameters, indoor environmental parameters and historical load data are preprocessed, including normalization, and the data are mapped to the [0,1] interval.

3. The chiller operation control method based on artificial intelligence according to claim 2 is characterized in that: The normalization process is as follows: For indoor environmental parameters Its normalized value for: Among them, min(x j ) and max(x j ) are the minimum and maximum values of the environmental parameter in the historical data respectively; For outdoor meteorological parameters Its normalized value for: in, and are the minimum and maximum values of the environmental parameter in the historical data respectively.

4. The chiller operation control method based on artificial intelligence according to claim 3 is characterized in that: For historical load data Q i,t；k , its normalized value for: Among them, min(Q) and max(Q) are the minimum and maximum values of the historical cooling load data of all rooms, respectively.

5. The chiller operation control method based on artificial intelligence according to claim 4 is characterized in that: The pre-processed indoor environmental parameters Outdoor meteorological parameters and historical load data Combined into input vector I i,t ,Right now: Then the input vector I i,t Input into the pre-trained long short-term memory network, i.e. LSTM model.

6. The chiller operation control method based on artificial intelligence according to claim 5 is characterized in that: The LSTM model consists of an input layer, multiple LSTM hidden layers, and an output layer; In the LSTM hidden layer, the calculation process of each LSTM unit is as follows: Input gate i t : i t =σ(W ii I i,t +W hi h t；1 +b i ) Forget Gate f t : f t =σ(W if I i,t +W hf h t；1 +b f ) Cell status update Cell State C t : Output gate o t :o t =σ(W io I i,t +W ho h t；1 +b o ) Hidden state h t : h t =o t ⊙tanh(C t ) Among them, σ is the sigmoid function, tanh is the hyperbolic tangent function W ii 、W hi 、W if 、W hf 、W ic 、W hc 、W io 、W ho is the weight matrix, b i 、b f 、b c 、b o is the bias vector, ⊙ represents element-by-element multiplication, h t；1 is the hidden state of the previous moment, C t；1 is the cell state at the previous moment; After calculations of multiple LSTM hidden layers, the output h of the last hidden layer is t Input to the output layer, and obtain the cooling load forecast value required by room i in the future preset time period Δt through linear transformation Among them, W out is the output layer weight matrix, b out is the output layer bias vector; Finally, the cooling load forecast values of all rooms are summarized to obtain the total cooling load forecast value required by the chiller system.

7. The chiller operation control method based on artificial intelligence according to claim 6, characterized in that: In S3, first, key features are extracted from the collected operating parameters of each component of the chiller. Assume that the chiller has m key operating parameters, and the value of the jth operating parameter at time t is recorded as x j,t , where j = 1, 2, ..., m; at the same time, record the cumulative running time T of the chiller from the time it was put into use to time t t , and the maintenance impact factor S quantified based on historical maintenance records t ; Use the multiple linear regression model to calculate the current performance attenuation coefficient α of the chiller t , considering the comprehensive impact of operating parameters, operating time and maintenance factors on unit performance degradation, its expression is: Among them, β0 is a constant term, β j is the regression coefficient, j=1,2,…,m+2; the regression coefficient can be obtained by training historical data;∈ t is a random error term, which obeys a normal distribution with a mean of 0; Use the least squares method to estimate the regression coefficient. Assume there are N sets of historical data. The goal is to minimize the sum of squared errors S: By calculating S with respect to β0, β1, ..., β m:2 By finding the partial derivative and setting it to 0, we can get a system of linear equations. Solving this system of equations can give us the estimated values of the regression coefficients. The initial performance coefficient curve of the chiller is represented by a polynomial function. Let the initial performance coefficient COP0 be the partial load rate PLR of the chiller, the chilled water supply temperature T chws , cooling water inlet temperature T cws The function is expressed as: COP0(PLR,T chws ,T cws ) <h2 style=";text-align:left;direction:ltr">= a0+a1PLR+a2(PLR)<h2 style=";text-align:left;direction:ltr"> 2 <h2 style=";text-align:left;direction:ltr"> +a3T<h2 style=";text-align:left;direction:ltr"> chws <h2 style=";text-align:left;direction:ltr"> +a4T<h2 style=";text-align:left;direction:ltr"> 2 <h2 style=";text-align:left;direction:ltr"> chws <h2 style=";text-align:left;direction:ltr"> +a5T<h2 style=";text-align:left;direction:ltr"> cws +a6T 2 cws +a7PLR·T chws +a8PLR·T cws +a9T chws ·T cws +a 10 PLR·T chws ·T cws Among them, a i are polynomial coefficients, i = 0, 1, ..., 10, obtained by fitting the factory test data or experimental data of the unit; Based on the calculated performance attenuation coefficient α t , correct the initial performance coefficient curve to obtain the actual performance coefficient curve COP at the current moment t (PLR,T chws ,T cws ), the correction formula is: COP t (PLR,T chws ,T cws )=α t ·COP0(PLR,T chws ,T cws ) Performance attenuation coefficient α t As a correction factor, the initial performance coefficient curve is corrected; To get the actual performance coefficient curve COP at the current moment t (PLR t ,T chws,t ,T cws,t ), according to the actual operating conditions of the chiller at the current moment, determine the key operating parameter value PLR,T corresponding to COP chws ,T cws , and substituting it into the corrected performance coefficient curve equation, we can get the current actual dynamic performance coefficient COP actual,t :COP actual,t =COP t (PLR t ,T chws,t ,T cws,t )=α t COP0(PLR t ,T chws,t ,T cws,t ).

8. The chiller operation control method based on artificial intelligence according to claim 7, characterized in that: In S4, the total energy consumption P of the air conditioning water system is minimized. total As the goal, while considering the total cooling load forecast value Demand; total energy consumption P total The energy consumption P of each chiller chiller , Energy consumption of chilled water pump P chwp , Energy consumption of cooling water pump P cwp , cooling tower fan energy consumption P tfan composition.

9. The chiller operation control method based on artificial intelligence according to claim 8, characterized in that: Total energy consumption of each chiller P chiller It can be expressed as: <h2 style=";text-align:left;direction:ltr">COP<h2 style=";text-align:left;direction:ltr"> i,0 <h2 style=";text-align:left;direction:ltr"> =a0+a1PLR<h2 style=";text-align:left;direction:ltr"> i <h2 style=";text-align:left;direction:ltr"> +a2(PLR<h2 style=";text-align:left;direction:ltr"> i <h2 style=";text-align:left;direction:ltr"> )<h2 style=";text-align:left;direction:ltr"> 2 <h2 style=";text-align:left;direction:ltr"> +a3T<h2 style=";text-align:left;direction:ltr"> chws <h2 style=";text-align:left;direction:ltr"> +a4T<h2 style=";text-align:left;direction:ltr"> 2 <h2 style=";text-align:left;direction:ltr"> chws <h2 style=";text-align:left;direction:ltr"> +a5T<h2 style=";text-align:left;direction:ltr"> cws +a6T 2 cws +a7PLR i ·T chws +a8PLR i ·T cws +a9T chws ·T cws +a 10 PLR i ·T chws ·T cws Among them, N chiler is the total number of chillers, P chiller,i is the power of the i-th chiller, Q chiller,i is the cooling capacity of the i-th chiller, COP i,0 is the COP value of the i-th chiller at the current moment obtained according to the initial performance curve, α t is the performance attenuation coefficient; the energy consumption of the chilled water pump P chwp and cooling water pump energy consumption P cwp and the speed of the water pump n pump Considering the operating efficiency of the water pump, according to the similarity law of the water pump, the energy consumption of the chilled water pump and the cooling water pump can be approximately expressed as: P chwp =b1·ω cHwp ·m cHw 2 +b2·ω cHwp 2 ·m cHw +b3·ω chwp 3 P cwp =c1·ω cwp ·m cw 2 +c2·ω cwp 2 ·m cw +c3·ω cwp 3 Among them, P chwp and P cwp are the power of the chilled water pump and the cooling water pump respectively, n chwp and n cwp is the speed of the chilled water pump and cooling water pump when they are running, n chwp,nom and n cwp,nom are the speeds of the chilled water pump and cooling water pump at rated operating conditions, m chw and m cw are the chilled water pump flow rate and the cooling water pump flow rate, ω chwp and ω cwp are the speed ratios of the chilled water pump and the cooling water pump respectively; Energy consumption of cooling tower P tfan The fan speed n of the cooling tower tfan The energy consumption expression can be obtained based on the empirical formula and experimental data fitting: P tfan =P tfan,nom (e0+e1PLR tfan +e2(PLR tfan ) 2 +e3(PLR tfan ) 3 ) Among them, P tfan is the actual power of the cooling tower fan, P tfan,nom is the rated power of the cooling tower fan, m ta is the actual air volume of the cooling tower, m ta,nom is the rated air volume of the cooling tower, e i is the model parameter, i = 0, 1, 2, 3, which can be obtained through the performance test of the cooling tower and data regression analysis; Then the objective function is: Then set constraint 1, cooling load constraint: the total cooling capacity Q provided by each chiller chiller The total cooling load forecast value must be met Right now: Among them, Q chiller,i is the cooling capacity of the i-th chiller in the current operating state; Set constraint 2, equipment operation range constraint: the start and stop status u of each chiller i ∈{0,1}; the speed of the chilled water pump n chwp The minimum speed n allowed chwp,min and maximum speed n chwp,max Between, that is, n chwp,min ≤n chwp ≤n chwp,max ; Cooling water pump speed n cwp Also, it must be at its minimum speed n cwp,min and maximum speed n cwp,max Between, that is, n cwp,min ≤n cwp≤ n cwp,max ; Cooling tower fan speed n tfan The minimum speed n must be met tfan,min and maximum speed n tfan,max The limit, that is, n tfan,min ≤n tfan≤ n tfan,max ; Set constraint 3, the chilled water supply temperature and cooling water inlet temperature range constraints of the chiller: chilled water supply temperature T chws and cooling water inlet temperature T cws Between their lowest and highest temperatures, respectively, T chws,min ≤T chws ≤T chws,max T cws,min ≤T cws ≤T cws,max Set constraint 4, flow range constraint of chilled water pump and cooling water pump: flow rate of each chilled water pump m chw,j and cooling water pump m cw,j The flow rate is between its minimum flow rate and maximum flow rate, that is, m chw,j,min ≤m chw,j ≤m chw,j,max m cw,j,min ≤m cw,j ≤m cw,j,max Set constraint 5, mutual constraints between devices: Among them, T chws and T chwr are the supply and return temperatures of chilled water, T cws and T cwr are the temperatures of cooling water entering and leaving the chiller, C pw is the specific heat of water. The genetic algorithm is used to solve the above objective function under the constraints: Initialize the population: Randomly generate a set of individuals that include the start and stop combinations of each chiller, the load rate of each chiller, the chilled water supply temperature, the cooling water inlet temperature, the chilled water pump speed, the cooling water pump speed, and the cooling tower fan speed to form the initial population; Fitness calculation: For each individual in the population, its fitness value is calculated according to the objective function. The smaller the fitness value, the better the individual. Selection operation: Roulette wheel selection method is used to select a certain number of individuals to enter the next generation according to their fitness values; Crossover operation: Perform crossover operation on the selected individuals to generate new individuals; Mutation operation: perform mutation operations on newly generated individuals to increase the diversity of the population; Termination condition judgment: When the preset termination condition of reaching the maximum number of iterations or the fitness value convergence is met, the iteration is stopped and the optimal individual is output. The start and stop combination of each chiller, the load rate of each chiller, the chilled water supply temperature, the cooling water inlet temperature, the chilled water pump speed, the cooling water pump speed, and the cooling tower fan speed corresponding to the optimal individual are the optimal operation control strategy.

10. The chiller operation control system based on artificial intelligence is characterized by: The method according to any one of claims 1 to 9 is adopted.

Images