CN-121978946-A - Greenhouse environment intelligent control system and control method based on multi-sensor fusion

CN121978946ACN 121978946 ACN121978946 ACN 121978946ACN-121978946-A

Abstract

The invention relates to the technical field of intelligent environmental control, in particular to a greenhouse environment intelligent control system and a control method based on multi-sensor fusion, wherein in the invention, a time sequence evolution standard is constructed by acquiring greenhouse multidimensional space-time environmental data and utilizing historical moment sensor data so as to generate an environment state dynamic coupling confidence coefficient; reconstructing a self-adaptive weighted fusion gain matrix to output a global environment representation value, and controlling a dead zone threshold value and a response lag time window by combining a deviation degree dynamic adjustment executing mechanism; the method realizes data verification and control strategy self-adaptive adjustment based on a time sequence rule, and effectively solves the problems of environment characterization distortion and frequent oscillation or response delay of equipment.

Inventors

Xuan Huaqiang
XIE ZHIGANG
LI HONGYU
GUO YUNPING
LIN GUOCHEN
GUO ZHENQUAN
Cai Murui

Assignees

聊城市农业科学院

Dates

Publication Date: 20260505
Application Date: 20260204

Claims (10)

1. The intelligent control method for the greenhouse environment based on the multi-sensor fusion is characterized by comprising the following steps of: S1, acquiring multidimensional space-time environmental data acquired by a distributed sensor array in a greenhouse, wherein the multidimensional space-time environmental data comprises current moment sensor data and stored historical moment sensor data; S2, constructing a time sequence evolution reference by utilizing the historical moment sensor data, and carrying out multidimensional consistency check on the current moment sensor data and the time sequence evolution reference to generate a current environment state dynamic coupling confidence coefficient; S3, dynamically reconstructing an adaptive weighted fusion gain matrix of the sensor array through a first mapping function by utilizing the environment state dynamic coupling confidence coefficient, and performing feature level fusion on the sensor data at the current moment by utilizing the adaptive weighted fusion gain matrix to output a global environment characterization value; S4, utilizing the environment state dynamic coupling confidence coefficient, combining the deviation degree of the sensor data at the current moment relative to the time sequence evolution reference, and dynamically adjusting an executing mechanism control dead zone threshold value and a response lag time window set by a proportional-integral-derivative controller through a second mapping function; s5, comparing the difference value with a preset target expected value according to the global environment representation value, and sending a control instruction to environment adjusting equipment of the greenhouse through the proportional-integral-derivative controller only when the difference value exceeds the dead zone threshold controlled by the executing mechanism and the duration exceeds the response lag time window.
2. The intelligent control method for the greenhouse environment based on the multi-sensor fusion according to claim 1, wherein the acquiring process of the multi-dimensional space-time environmental data specifically comprises the following steps: deploying a greenhouse internal distributed sensor array in the greenhouse according to the three-dimensional space gridding node position, wherein each sensor node in the greenhouse internal distributed sensor array is configured to synchronously acquire air temperature, air humidity, illumination intensity and carbon dioxide concentration parameters; The method comprises the steps of reading output values of a sensor array in a current sampling period in real time through a field bus communication protocol, performing time stamp alignment on read data by adopting a nearest neighbor interpolation method, filling missing values by using an effective value at the last time, and constructing sensor data at the current time; the method comprises the steps of calling a parameter set in a continuous time sequence from a last sampling period to a current sampling period of a distributed sensor array in a greenhouse recorded in a local storage module, and marking the parameter set as stored historical moment sensor data; and carrying out association integration on the current time sensor data and the historical time sensor data according to space coordinates and time stamps to obtain multidimensional space-time environmental data.
3. The intelligent control method for the greenhouse environment based on the multi-sensor fusion according to claim 1, wherein the construction process of the time sequence evolution standard specifically comprises the following steps: Reading a parameter sequence in a latest continuous time window in the historical moment sensor data, and respectively performing least square linear fitting on parameters of air temperature, air humidity, illumination intensity and carbon dioxide concentration of each node in the distributed sensor array in the greenhouse; And calculating a theoretical predicted value and an allowable fluctuation interval at the current moment according to a change slope obtained by least square linear fitting, and establishing a parameter set containing the theoretical predicted value and the allowable fluctuation interval as a time sequence evolution reference.
4. The intelligent control method for the greenhouse environment based on the multi-sensor fusion according to claim 1, wherein the process of generating the environment state dynamic coupling confidence coefficient specifically comprises the following steps: Calculating the absolute value of the difference between the actual acquired value of each sensor node in the sensor data at the current moment and the corresponding theoretical predicted value in the time sequence evolution reference; Mapping the absolute value of the difference value into a normalized value between zero and one by using an inverse proportion function, wherein the normalized value represents the single-point consistency degree; And calculating an arithmetic average value of the normalized values calculated by all the nodes, and taking the arithmetic average value as an environment state dynamic coupling confidence coefficient.
5. The intelligent control method for the greenhouse environment based on the multi-sensor fusion according to claim 1, wherein the generation process of the adaptive weighted fusion gain matrix specifically comprises the following steps: Invoking a locally stored preset first mapping function configured as an exponential decay model with an environmental state dynamic coupling confidence coefficient as an argument; Inputting the environment state dynamic coupling confidence coefficient into the first mapping function, and calculating a confidence weight adjustment factor for each sensor node in the distributed sensor array in the greenhouse; calculating local weight values of all the sensor nodes according to the credibility weight adjusting factors, and executing normalization processing on all the local weight values; Constructing a diagonal matrix with normalized local weight values as diagonal elements, and establishing the diagonal matrix as an adaptive weighted fusion gain matrix.
6. The intelligent control method for the greenhouse environment based on the multi-sensor fusion according to claim 5, wherein the generation process of the global environment characterization value specifically comprises the following steps: The values of the air temperature, the air humidity, the illumination intensity and the carbon dioxide concentration contained in the sensor data at the current moment are arranged according to the serial numbers of the sensor nodes, and an observation feature vector in a column vector form is constructed; performing matrix multiplication operation by utilizing the self-adaptive weighted fusion gain matrix to multiply the observation feature vector left so as to carry out weighted correction on the data acquired by each sensor node, and generating a weighted environment parameter vector; And carrying out accumulation and summation operation on all dimension elements in the weighted environmental parameter vector, calculating to obtain a single numerical value reflecting the overall environmental condition of the greenhouse, and marking the numerical value as a global environmental characterization value and outputting the global environmental characterization value.
7. The intelligent control method for the greenhouse environment based on the multi-sensor fusion according to claim 1, wherein the process of controlling the dead zone threshold and the response lag time window by the dynamic adjustment executing mechanism in S4 specifically comprises: calculating the sum of absolute values of differences between all node values in the sensor data at the current moment and corresponding values in the time sequence evolution reference, and determining the sum as the deviation degree; invoking an internally stored second mapping function configured to dynamically couple a product of an inverse of the confidence coefficient and the degree of deviation with the environmental state as computational logic; And executing the operation process to obtain a target control parameter value, and directly writing the target control parameter value into a parameter register of the proportional-integral-derivative controller for simultaneously updating the dead zone threshold value and the response lag time window of the actuator.
8. The intelligent control method for the greenhouse environment based on the multi-sensor fusion according to claim 1, wherein the sending process of the control instruction specifically comprises the following steps: Calculating the absolute value of the difference between the global environment representation value and a preset target expected value; Comparing the absolute value of the difference value with a currently set dead zone threshold controlled by an actuating mechanism in real time; when the absolute value of the difference is larger than the dead zone threshold controlled by the executing mechanism, starting an internal timer to perform continuous timing monitoring; And if the timer confirms that the state that the absolute value of the difference value is larger than the dead zone threshold controlled by the executing mechanism continuously exists and the accumulated time length exceeds the response lag time window, calculating based on the absolute value of the difference value by utilizing the proportional-integral-derivative controller to generate a control instruction, and sending the control instruction to environment regulating equipment of the greenhouse through a hardware interface so as to execute regulating action.
9. The intelligent control method for greenhouse environment based on multi-sensor fusion according to claim 3, wherein the calculation process of the theoretical predicted value specifically comprises the following steps: Based on a parameter sequence in a latest continuous time window in the sensor data of the historical moment, aiming at each type of environmental parameter of each sensor node, obtaining a change slope and an intercept which uniquely represent a change trend through least square linear fitting calculation, and calculating a theoretical predicted value of the parameter at the current moment according to the change slope and the intercept.
10. A system for using a greenhouse environment intelligent control method based on multi-sensor fusion according to any one of claims 1-9, comprising a multi-dimensional spatiotemporal data acquisition module (100), an environment state verification and confidence assessment module (200), an adaptive weighted feature fusion module (300), a control strategy parameter dynamic adjustment module (400) and a feedback decision and execution module (500), wherein: The multi-dimensional space-time data acquisition module (100) acquires multi-dimensional space-time environment data acquired by a distributed sensor array in a greenhouse, wherein the multi-dimensional space-time environment data comprises current time sensor data and stored historical time sensor data; The environmental state verification and confidence evaluation module (200) utilizes the historical moment sensor data to construct a time sequence evolution reference, and carries out multidimensional consistency verification on the current moment sensor data and the time sequence evolution reference to generate a current environmental state dynamic coupling confidence coefficient; The self-adaptive weighted feature fusion module (300) dynamically reconstructs a self-adaptive weighted fusion gain matrix of the sensor array through a first mapping function by utilizing the environment state dynamic coupling confidence coefficient, and performs feature level fusion on the sensor data at the current moment by utilizing the self-adaptive weighted fusion gain matrix to output a global environment characterization value; the control strategy parameter dynamic adjustment module (400) dynamically couples the confidence coefficient by utilizing the environment state, combines the deviation degree of the sensor data at the current moment relative to the time sequence evolution reference, and dynamically adjusts an executing mechanism control dead zone threshold value and a response lag time window set by a proportional-integral-derivative controller through a second mapping function; and the feedback decision and execution module (500) compares the difference value with a preset target expected value according to the global environment representation value, and only when the difference value exceeds the dead zone threshold controlled by the execution mechanism, and after the duration exceeds the response lag time window, a control instruction is sent to the environment regulating equipment of the greenhouse through the proportional-integral-derivative controller.

Description

Greenhouse environment intelligent control system and control method based on multi-sensor fusion Technical Field The invention relates to the technical field of intelligent environmental control, in particular to a greenhouse environment intelligent control system and a control method based on multi-sensor fusion. Background As modern facility agriculture evolves towards a high definition, greenhouse environment monitoring systems have widely adopted distributed sensor networks to sense complex microclimate changes. The greenhouse environment is essentially a non-stationary random system with large time lags, strong nonlinearities and high coupling characteristics, and the thermodynamic field and gas concentration field distribution inside the system often show significant space-time dynamic change characteristics. In prior art systems, for multidimensional environmental data acquired from a distributed array, conventional processing logic tends to focus on cross-sectional statistical property analysis or transient state-based linear weighting of the data at the same time, ignoring the depth correlation of the evolution law of sensor data in the continuous time dimension with the historical reference. This means that when local non-uniformity turbulence occurs in the greenhouse or weak non-monotonic drift occurs in the sensor itself, the system is difficult to establish an effectiveness checking mechanism based on a time sequence evolution rule, so that the dynamic confidence level of the current data cannot be accurately estimated, and the generated environment characterization value may contain false trend information. In addition, at the control strategy execution level, the action of the environment adjusting mechanism (such as an electric sunshade net, a wet curtain fan and the like) is generally taken over by a proportional-integral-derivative controller in cooperation with preset trigger logic. However, the control dead zone threshold and the response lag time window in the prior art are usually set to constant values in the system initialization stage, the static parameter configuration mode breaks through the internal connection between the data acquisition quality and the control execution logic, when the environment is in a high-frequency fluctuation state or the confidence of sensor data is reduced, the rigid control threshold lacks the capability of self-adaptive expansion and contraction according to the environment deviation degree, the control response mechanism is extremely easy to cause frequent misoperation and invalid switching of an executing mechanism in a critical state, the vibration not only aggravates the fatigue loss of mechanical parts and causes invalid dissipation of energy sources, but also is difficult to maintain long-term steady-state convergence of an internal environment field of a greenhouse, and the double severe requirements of modern agriculture on steady-state maintenance and accurate response of the growing environment cannot be met. Disclosure of Invention The invention aims to provide a greenhouse environment intelligent control system and a control method based on multi-sensor fusion, which are used for solving the problems in the background technology. The specific technical problems include how to construct a time sequence evolution reference by utilizing historical sensor data to generate an environmental state dynamic coupling confidence coefficient, and accordingly, the self-adaptive weighted fusion of the sensor data and the dynamic adjustment of an execution mechanism control dead zone threshold value and a response lag time window are realized, so as to solve the technical problems of environmental characterization distortion caused by lack of verification of a data time sequence evolution rule and frequent oscillation or response lag of the execution mechanism caused by incapability of adapting to environmental fluctuation confidence coefficient by static control logic in the prior art. In order to achieve the above purpose, one of the purposes of the invention is to provide a greenhouse environment intelligent control method based on multi-sensor fusion, comprising the following steps: S1, acquiring multidimensional space-time environmental data acquired by a distributed sensor array in a greenhouse, wherein the multidimensional space-time environmental data comprises current moment sensor data and stored historical moment sensor data, and specifically comprises the following steps: Deploying a greenhouse internal distributed sensor array in the greenhouse according to the three-dimensional space gridding node position, wherein each sensor node in the greenhouse internal distributed sensor array is configured to synchronously acquire air temperature, air humidity, illumination intensity and carbon dioxide concentration parameters; The method comprises the steps of reading output values of a sensor array in a current sampling period in real time through