An intelligent feeding control method and system based on big data breeding

Abstract

This invention relates to the field of aquaculture feeding technology, specifically to an intelligent feeding control method and system based on big data aquaculture. The method includes: if the pause judgment index value at the t-th judgment time is less than a preset coefficient threshold, then feed feeding is paused; if the pause judgment index value at the (t+1)-th judgment time is not less than the preset coefficient threshold, then feed feeding resumes at the (t+2)-th judgment time, and it is determined whether the feeding rate needs to be adjusted. If adjustment is needed, then the feed feeding rate at the (t+3)-th judgment time is calculated based on the feed feeding rate at the (t+2)-th judgment time and the trend value of the proportional coefficient; if not, then the feed feeding rate at the (t+2)-th judgment time is used as the feed feeding rate at the (t+3)-th judgment time, and the pause judgment index value at the (t+3)-th judgment time is continuously acquired until the feeding termination condition is triggered, at which point feed feeding ends. This invention can improve the control effect of aquaculture feeding.

Description

Technical Field

[0001] This invention relates to the field of aquaculture and feeding technology, specifically to an intelligent feeding control method and system based on big data aquaculture. Background Technology

[0002] In the aquaculture industry, feed control is a key measure to optimize animal growth, health and aquaculture efficiency. The rationality of the feeding strategy directly determines the growth rate, feed conversion rate and overall economic benefits of the farmed animals. Therefore, how to control feed is crucial to the health of farmed animals and the aquaculture efficiency.

[0003] Current automated aquaculture feeding control generally adopts a preset fixed-point, fixed-quantity feeding mode, such as feeding a fixed amount of feed at fixed time periods. However, this fixed-point, fixed-quantity feeding mode does not fully consider the feeding behavior and actual needs of the farmed animals, which can easily lead to problems such as insufficient or excessive feeding in certain areas. This means that feed may accumulate in areas where the farmed animals are not very active, resulting in waste, or insufficient feeding may occur in areas where the farmed animals are very active. Insufficient or excessive feeding in certain areas will lead to feed waste and reduced farming efficiency. Therefore, how to adaptively adjust the feeding strategy to solve the problems of feed waste and reduced farming efficiency caused by insufficient or excessive feeding has become an urgent problem to be solved. Summary of the Invention

[0004] To address the aforementioned problems, this invention provides an intelligent feeding control method and system based on big data aquaculture, the specific technical solution of which is as follows:

[0005] In a first aspect, embodiments of the present invention provide an intelligent feeding control method based on big data aquaculture, comprising the following steps:

[0006] Obtain the pause judgment index value at the t-th judgment time of the current feed dispensing cycle at the target feed dispensing point. Determine if the pause judgment index value at the t-th judgment time is less than a preset coefficient threshold. If it is less, control the target feed dispensing point to pause feed dispensing until the (t+1)-th judgment time. Continue to obtain the pause judgment index value at the (t+1)-th judgment time of the current feed dispensing cycle. Determine if the pause judgment index value at the (t+1)-th judgment time is less than a preset coefficient threshold. If it is not less, restore the feed dispensing rate at the (t+2)-th judgment time to the feed dispensing rate before the pause, and calculate the proportional coefficient trend value at the (t+2)-th judgment time. The trend value is used to determine whether the feed delivery rate needs to be adjusted. If adjustment is needed, the feed delivery rate at the (t+2)th judgment time is calculated based on the trend value of the feed delivery rate and the proportional coefficient at the (t+2)th judgment time. The feed delivery rate at the target feed delivery point is then controlled to be the same as that at the (t+3)th judgment time from the (t+2)th judgment time to the (t+3)th judgment time. If no adjustment is needed, the feed delivery rate at the (t+2)th judgment time is used as the feed delivery rate at the (t+3)th judgment time. The pause judgment index value at the (t+3)th judgment time of the current feed delivery cycle is continuously obtained until the feed delivery end condition is triggered, at which point the feed delivery to the target feed delivery point in the current feed delivery cycle is terminated.

[0007] Secondly, embodiments of the present invention provide an intelligent feeding control system for big data-based aquaculture, including a memory and a processor, wherein the processor executes a computer program stored in the memory to implement the aforementioned intelligent feeding control method for big data-based aquaculture.

[0008] Beneficial effects: First, obtain the pause judgment index value at the t-th judgment time of the current feed dispensing cycle at the target feed dispensing point. Determine if the pause judgment index value at the t-th judgment time is less than a preset coefficient threshold. If it is less, control the target feed dispensing point to pause feed dispensing until the (t+1)-th judgment time. Continue to obtain the pause judgment index value at the (t+1)-th judgment time of the current feed dispensing cycle. Determine if the pause judgment index value at the (t+1)-th judgment time is less than a preset coefficient threshold. If it is not less, restore the feed dispensing rate at the (t+2)-th judgment time to the feed dispensing rate before the pause, and calculate the proportional coefficient trend value at the (t+2)-th judgment time. Based on the (t+2)-th judgment time... The feed distribution rate trend value is used to determine whether an adjustment is needed. If adjustment is needed, the feed distribution rate at the (t+2)th judgment time is calculated based on the feed distribution rate and the feed distribution rate trend value at the (t+3)th judgment time. The feed distribution rate at the target feed distribution point is then controlled to be the same from the (t+2)th judgment time to the (t+3)th judgment time. If no adjustment is needed, the feed distribution rate at the (t+2)th judgment time is used as the feed distribution rate at the (t+3)th judgment time. The pause judgment index value at the (t+3)th judgment time of the current feed distribution cycle is continuously acquired until the feed distribution end condition is triggered, at which point the feed distribution to the target feed distribution point for the current feed distribution cycle ends. This invention uses strategies such as feed distribution pause and feed distribution rate adjustment to control the feeding of the target feed distribution point in the current feed distribution cycle. This can minimize the inflexibility of fixed-point and quantitative feeding methods and minimize local underfeeding or overfeeding, thereby improving breeding efficiency and feed utilization. Attached Figure Description

[0009] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a flowchart of an intelligent feeding control method for aquaculture based on big data, according to the present invention. Detailed Implementation

[0011] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention are within the protection scope of the embodiments of the present invention.

[0012] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art.

[0013] This embodiment provides an intelligent feeding control method based on big data aquaculture, which is described in detail below:

[0014] like Figure 1 As shown, this intelligent feeding control method for aquaculture based on big data includes the following steps:

[0015] Step S001: Obtain the pause judgment index value at the t-th judgment time of the current feed delivery cycle of the target feed delivery point, and determine whether the pause judgment index value at the t-th judgment time is less than the preset coefficient threshold. If it is less, control the target feed delivery point to pause feed delivery until the (t+1)-th judgment time.

[0016] This embodiment addresses the lack of flexibility in fixed-point, quantitative feeding methods by adaptively adjusting the feeding rate based on the actual activity status of the animals within the breeding area before feeding, or by avoiding feed waste and decreased breeding efficiency caused by localized underfeeding or overfeeding. Specifically, this embodiment controls feed distribution or feeding during breeding by analyzing feed usage, feed usage trends, adjusting feed dispensing speed, pausing feed dispensing, and monitoring animal activity. In other words, this embodiment, through adaptive adjustment of the feeding strategy, can solve or minimize the occurrence of underfeeding or overfeeding, thus avoiding feed waste and decreased breeding efficiency. Since there are generally multiple feed delivery points or feed delivery locations within a breeding area or farm, and since the feeding control process or feed delivery control process is the same for each feed delivery point or feed delivery location within the same breeding area or farm in this embodiment, for ease of understanding and description, this embodiment will subsequently describe the feed delivery process of any feed delivery point in any breeding area where the breeding object is chicken as an example, and the breeding area will be referred to as the target breeding area, and the feed delivery point will be referred to as the target feed delivery point.

[0017] Furthermore, since feeding or feed delivery is not continuous during aquaculture, it is usually done on a timed basis, with each feeding or delivery having a specific time requirement. This embodiment will subsequently describe the process of an upcoming feeding or delivery as an example, and record it as the current feeding cycle. If the total duration of the upcoming feeding or delivery is 30 minutes, and the start time of the upcoming feeding or delivery is r, then the 30 minutes after time r is the current feeding cycle. Since this embodiment describes the feeding process at the target feeding point as an example, the location of feeding within the current feeding cycle in this embodiment is mainly the target feeding point.

[0018] After determining the current feeding cycle, the feeding rate at the target feeding point at the start time of the current feeding cycle is obtained based on the activity of the farmed organisms in the target breeding area before the start of the current feeding cycle. In this embodiment, the start time of the current feeding cycle is the beginning time of the current feeding cycle. Obtaining the feeding rate at the start time of the current feeding cycle based on the activity of the farmed organisms in the target breeding area before the start of the current feeding cycle is a fundamental measure to avoid overfeeding or underfeeding at the end of the current feeding cycle. The specific process for obtaining the feeding rate at the start time of the current feeding cycle is as follows:

[0019] First, a preset monitoring time period before the start of the current feed delivery cycle is obtained, and a high-definition camera is used to collect video of the target breeding area within the preset monitoring time period. That is, the time period corresponding to the video is the preset monitoring time period, and the area corresponding to the video is the target breeding area. In this embodiment, the actual area corresponding to each frame of the acquired video is consistent, and the size of each frame is consistent. That is, the physical space covered by each frame of the video is exactly the same. The camera is fixed and the viewing angle is stable. In other words, the physical location of the same coordinate pixels in different images of the acquired video is the same in the target breeding area. Moreover, since the activity of the breeding objects is random, the activity of the breeding objects within a certain period of time before feeding can be monitored and analyzed more accurately to analyze the activity of the breeding objects in different locations or areas of the breeding area. In specific applications, the implementer can set the preset monitoring time period according to the actual situation such as the activity of the breeding objects and the accuracy of monitoring the activity of the breeding objects. For example, two hours before the start of the current feed delivery cycle can be used as the preset monitoring time period.

[0020] Then, using a preset grid, each frame of the video is divided into non-overlapping regions to obtain region blocks for each frame. The specific process is as follows: The size of the preset grid is denoted as C×C, where C is a positive integer. It is determined whether the width and height of the image are both multiples of C. If so, each frame is evenly divided into... If not, then divide each frame of image evenly into regions. The image is divided into regions, and each region in each frame is recorded as a region block in the corresponding image. The image uses a rounding down sign, H0 as the image width, and W0 as the image height. The preset grid has a height and width of C, and each region corresponds to a physical sub-region within the target breeding area. In practical applications, the implementer can set the value of C according to the actual situation of the breeding object, such as choosing a smaller grid size for smaller breeding objects and a larger grid size for larger breeding objects. In this embodiment, the breeding object is chicken, which is relatively small, so the value of C can be set smaller, such as 50 or 100 pixels. Simultaneously, the YOLO model is used to identify and detect the chickens in each frame of the image, obtaining the real-time activity position of the breeding object in each frame, and counting the number of breeding objects in each region of each frame. For example, the collected images are input into the trained YOLO model, which outputs the detection boxes for chickens in each frame, counts the number of detection box center points falling in each region, and records this as the number of breeding objects in the corresponding region. The training process of the YOLO model is a well-known technique, and the YOLO model is an efficient object detection algorithm.

[0021] Next, by analyzing the number of farmed animals in consecutive blocks within the same area of ​​the target aquaculture area during the preset monitoring time period and the changes in the number of farmed animals, the activity frequency of farmed animals in the sub-regions of the aquaculture area is analyzed, which is the activity frequency assessment value. Subsequently, the feed delivery rate at the start of feed delivery will be determined by the obtained activity frequency assessment value. Based on the above description, this embodiment needs to first obtain areas in the target aquaculture area that have a physical spatial mapping relationship with the area blocks in the video images, and record them as sub-regions in the target aquaculture area. Since the pixels with the same coordinates in different images of the video obtained in this embodiment correspond to the same physical location in the target aquaculture area, the area blocks at the same location in different images of the video are all the same part of the aquaculture area. That is to say, dividing the image is equivalent to dividing the actual spatial target aquaculture area. Then, in all the area blocks in the images of the video, all the area blocks that have a physical spatial mapping relationship with each sub-region are obtained, and all the area blocks that have a physical spatial mapping relationship with each sub-region are arranged according to the order of image acquisition. The result of the arrangement is recorded as the area block sequence of the corresponding sub-region. If a certain area block represents the actual target aquaculture area If the range in the image is exactly a sub-region within the aquaculture area, then there is a physical spatial mapping relationship between this region block and the sub-region, or vice versa. Furthermore, for any sequence of region blocks within a sub-region, the x-th region block in the sequence belongs to the x-th frame of the video, and each region block in the sequence is part of the image content of that sub-region. In other words, the center position of all region blocks in the sequence is the same within the sub-region. Then, based on the number of aquaculture objects in each region block of the sequence within the aquaculture area and the image to which the region block belongs... The time interval between the collection time and the start time of feed feeding in the current feed feeding cycle is used to obtain the activity frequency assessment value corresponding to each sub-region in the breeding area. The relative change in the number of farmed objects between adjacent regional blocks in the regional block sequence is the ratio of the result of positively correlated mapping of the number of farmed objects in the next regional block to the result of positively correlated mapping of the number of farmed objects in the previous regional block. For ease of understanding, this embodiment will subsequently describe the specific process of obtaining the activity frequency assessment value corresponding to any sub-region A in the breeding area as an example. The specific process of obtaining the activity frequency assessment value corresponding to sub-region A is as follows:

[0022] Based on the number of farmed animals in each block of the sub-region A's block sequence, the changes in the number of farmed animals between adjacent blocks in the sub-region A's block sequence, and the time interval between the acquisition time of the image to which each block belongs in the sub-region A's block sequence belongs and the start time of feed feeding in the current feed feeding cycle, a weighted frequent representation value sequence corresponding to sub-region A is obtained. The a-th weighted frequent representation value in the weighted frequent representation value sequence corresponding to sub-region A is the result of multiplying the normalized result of the number of farmed animals in the (a+1)-th block of the sub-region A's block sequence, the relative change in the number of farmed animals in the (a+1)-th block, and the distance factor of the (a+1)-th block. The number of weighted frequent representation values ​​in the weighted frequent representation value sequence corresponding to sub-region A is the total number of blocks in the sub-region A's block sequence minus 1. The relative change in the number of farmed animals in the (a+1)-th block is the number of farmed animals in the (a+1)-th block of the sub-region A's block sequence. The ratio of the positive correlation mapping result to the positive correlation mapping result of the number of farmed objects in the a-th region block. The greater the relative change in the number of farmed objects in the (a+1)-th region block, the more frequent the activity of farmed objects in the (a+1)-th region block is compared to the a-th region block. The distance factor of the (a+1)-th region block is the result of reverse probability normalization of the time interval between the acquisition time of the image to which the (a+1)-th region block belongs and the start time of the feed feeding in the current feed feeding cycle. Here, normalization is implemented using the normalization function Norm(). Positive correlation mapping is implemented using an exponential function with a constant e as the base. Reverse probability normalization refers to performing probability normalization after negative correlation mapping. Negative correlation mapping is implemented using a negative exponential function with a constant e as the base. The result of summing all weighted frequent representation values ​​in the weighted frequent representation value sequence corresponding to sub-region A is denoted as the activity frequency evaluation value corresponding to sub-region A; and the specific expression of the activity frequency evaluation value corresponding to sub-region A is:

[0023]

[0024]

[0025]

[0026] in, Let M0 be the activity frequency assessment value corresponding to sub-region A, and M0 be the total number of weighted frequent representation values ​​in the weighted frequent representation value sequence corresponding to sub-region A. M0 is also the total number of region blocks in the region block sequence of sub-region A minus 1. Let be the a-th weighted frequent representation value in the weighted frequent representation value sequence corresponding to sub-region A. , Let be the unweighted value of the a-th weighted frequent representation in the weighted frequent representation sequence corresponding to subregion A. Norm() is the normalization function, and exp() is the exponential function with base e. This represents the number of farmed objects in the (a+1)th block of the block sequence in sub-region A. Let be the number of farmed objects in the a-th block of the block sequence in sub-region A. , To represent the relative change in the number of aquaculture species in the (a+1)th region, an exponential function with base e is used to... and Performing a positive correlation mapping is to avoid a denominator of 0 when directly calculating the ratio, thus making the calculation meaningful. Let M1 be the distance factor of the (a+1)th region block, and M1 be the total number of region blocks in the region block sequence of sub-region A. The time interval between the acquisition time of the image of the (a+1)th region block and the start time of feed distribution in the current feed distribution cycle; and the greater the number of farmed animals in the region block sequence of sub-region A, and the greater the relative change in the number of farmed animals in the region block sequence of sub-region A, the greater the number of farmed animals in sub-region A, and the more frequent their activities may be. In addition, the features of the region blocks that are closer to the start time of feed distribution in the current feed distribution cycle are more reflective of the situation of farmed animals in sub-region A in the current feed distribution cycle. Therefore, this embodiment makes the number of farmed animals and the relative change in the number of farmed animals in the region blocks that are closer to the start time of feed distribution in the current feed distribution cycle participate more or dominate more in calculating the activity frequency assessment value corresponding to sub-region A. and The larger, or rather A larger value indicates a greater number of farmed organisms in sub-region A and potentially more frequent activity, or a greater number of farmed organisms in sub-region A during the current feed application cycle and potentially more frequent activity. The smaller the value, the fewer the number of farmed organisms in sub-region A, and the less frequent their activity is likely to be. When it is larger, The larger; therefore A larger value indicates that there are more farmed animals in sub-region A and their activities are more frequent, or that there may be more farmed animals in sub-region A and their activities may be more frequent during the current feeding cycle. When there may be more farmed animals in sub-region A and their activities may be more frequent during the current feeding cycle, it indicates that the potential feeding demand of farmed animals in sub-region A is greater. In order to avoid the problem of insufficient feeding at the end of the current feeding cycle, the feed feeding rate at the target feed feeding point at the beginning of the current feeding cycle should be larger. The smaller the value, the fewer the number of farmed animals in sub-region A and the less frequent their activity. Alternatively, it may indicate that the number of farmed animals in sub-region A during the current feeding cycle is likely to be small and their activity may be relatively infrequent. When the number of farmed animals in sub-region A during the current feeding cycle is likely to be small and their activity may be relatively infrequent, it indicates that the potential feeding demand of farmed animals in sub-region A is relatively small. To avoid overfeeding at the end of the current feeding cycle, the feed delivery rate to the target feed delivery point should be relatively small at the beginning of the current feeding cycle.

[0027] Since the activity frequency assessment value is related to the feed delivery rate at the start of the current feed delivery cycle, after obtaining the activity frequency assessment value corresponding to the sub-region, this embodiment obtains the feed delivery rate at the start of the current feed delivery cycle based on the activity frequency assessment values ​​corresponding to all sub-regions, the distance between each sub-region and the target feed delivery point, and the preset feed delivery rate requirement range. The sub-regions closer to the target feed delivery point better reflect the situation of the farmed animals near the target feed delivery point. The preset feed delivery rate requirement range is the value range of the initial delivery rate and is also a basic measure to avoid underfeeding and overfeeding. The specific process of obtaining the feed delivery rate at the start of the current feed delivery cycle based on the activity frequency assessment values ​​corresponding to all sub-regions, the distance between each sub-region and the target feed delivery point, and the preset feed delivery rate requirement range is as follows:

[0028] In actual feeding, since feed distribution points typically only effectively attract and supply livestock within a certain surrounding area, the impact of livestock activity in different sub-regions on the same distribution point is closely related to its spatial distance. Furthermore, if livestock activity is relatively low in the area surrounding a feed distribution point, maintaining a large feed intake can easily lead to feed accumulation and waste. Conversely, if livestock activity is frequent and highly concentrated in the area surrounding the distribution point, insufficient feed intake can intensify local competition for food, affecting the balanced feeding and uniform growth of the livestock. Therefore, after obtaining the activity frequency assessment value for the corresponding sub-region, it is necessary to comprehensively consider the target feed distribution. The activity frequency and spatial distribution of aquatic organisms in the surrounding sub-regions are analyzed to quantitatively assess the initial feed demand at the target feed delivery point. Specifically, the initial feed demand assessment value for the target feed delivery point is obtained based on the activity frequency assessment values ​​for all sub-regions and the distance between each sub-region and the target feed delivery point. This initial feed demand assessment value reflects the level of feed demand at the target feed delivery point at the start of the current feed delivery cycle. Then, based on the initial feed demand assessment value and the preset feed delivery rate requirement range, the feed delivery rate at the start of the current feed delivery cycle is obtained.

[0029] The specific process for obtaining the initial feed demand assessment value of the target feed delivery point based on the activity frequency assessment values ​​of all sub-regions and the distance between each sub-region and the target feed delivery point is as follows:

[0030] The distances between each sub-region and the target feed delivery point are obtained, and these distances are then subjected to inverse probability normalization. The final result is recorded as the weight factor for the corresponding sub-region. Inverse probability normalization involves first performing a negative correlation mapping using a negative exponential function with a base of e, followed by probability normalization. The distance between each sub-region and the target feed delivery point is the Euclidean distance between the center point of the corresponding sub-region and the target feed delivery point. The weight factor for the j-th sub-region in the target aquaculture area is... , Let J be the distance between the j-th sub-region and the target feed delivery point, and J be the total number of sub-regions in the target aquaculture area. Calculate the product of the activity frequency assessment value corresponding to each sub-region and the corresponding weight factor, and record it as the weighted activity frequency assessment value corresponding to the corresponding sub-region. Accumulate the weighted activity frequency assessment values ​​corresponding to all sub-regions and record it as the initial feed demand assessment value of the target feed delivery point. The expression for the initial feed demand assessment value of the target feed delivery point is:

[0031]

[0032] Where H represents the initial feed demand assessment value at the target feed delivery point, and J represents the total number of sub-regions within the target aquaculture area. This represents the activity frequency assessment value corresponding to the j-th sub-region. Let be the weighting factor for the j-th sub-region; and the closer the j-th sub-region is to the target feed delivery point, the more accurately the activity of farmed animals in the j-th sub-region reflects the feeding demand level of farmed animals around the target feed delivery point. In other words, the closer the sub-region is to the target feed delivery point, the greater its contribution to the initial feed demand assessment value. The initial feed demand assessment value reflects the feed delivery rate at the start of feed delivery. Furthermore, the more farmed animals and the more frequently they move in the j-th sub-region before the start of feed delivery, and the closer the j-th sub-region is to the target feed delivery point, the greater its contribution to the initial feed demand assessment value. The higher the initial feed demand, the greater the feed requirement at the target feed point at the start of the current feed cycle. To avoid insufficient feeding at the end of the current feed cycle, the feed delivery rate to the target feed point should be higher at the beginning of the current feed cycle if the initial demand is higher. The larger H is, the greater the demand for feed at the target feed point at the beginning of the current feed feeding cycle. Therefore, the feed feeding rate at the beginning of the current feed feeding cycle should be higher. Conversely, the smaller H is, the smaller the demand for feed at the target feed point at the beginning of the current feed feeding cycle. To avoid overfeeding, the feed feeding rate at the target feed point at the beginning of the current feed feeding cycle should be lower.

[0033] The specific process of obtaining the feed delivery rate at the start of the current feed delivery cycle based on the initial feed demand assessment value of the target feed delivery point and the preset feed delivery rate requirement range is as follows: First, obtain the preset feed delivery rate requirement range. The preset feed delivery rate requirement range is also a basic measure to avoid underfeeding or overfeeding. The maximum value in the preset feed delivery rate requirement range is the maximum feed delivery rate allowed by the feeding or feeding equipment applied to the target feed delivery point. The minimum value in the preset feed delivery rate requirement range is set based on historical feeding data. For example, the product of the average feed consumption rate of the target feed delivery point in the previous feed delivery cycle and the preset proportional coefficient can be used as the minimum value in the preset feed delivery rate requirement range. The preset proportional coefficient can be adjusted and set according to the actual situation such as the stocking density to avoid the feed delivery rate being too low and affecting the feeding attractiveness of the farmed animals, while preventing overfeeding and feed waste, thereby achieving a balance between ensuring feeding demand and improving feed utilization. However, its value is required to be within the range of 0 to 1. For example, in this embodiment, 0.6 can be used as the preset proportional coefficient. Then, the range of the preset feed delivery rate requirement interval is multiplied by the normalized result of the initial feed demand assessment value, and the result is added to the minimum value of the preset feed delivery rate requirement interval. This result is denoted as the feed delivery rate at the start of the current feed delivery cycle. In other words, the feed delivery rate at the start of the current feed delivery cycle is... ,in, This is the minimum value within the preset feed delivery rate requirement range. H represents the maximum value within the preset feed delivery rate requirement range, where H is the initial feed demand assessment value at the target feed delivery point. The minimum value within the range of initial feed requirement assessment values. The maximum value within the range of initial feed requirement assessment values. Let H be the expression for minimax normalization, and This part aims to ensure that the feed infeed rate at the start of the feeding cycle is within the preset feed infeed rate requirement range. A larger H indicates a higher feed infeed rate at the start of the current feeding cycle, and vice versa. Furthermore, this embodiment determines the feed infeed rate at the start of the current feeding cycle based on the preset feed infeed rate requirement range and the initial feed demand assessment value at the target feed infeeding point. This can avoid or reduce the probability of underfeeding or overfeeding, thereby improving breeding efficiency and feed utilization.

[0034] Therefore, this embodiment obtains the feed dispensing rate at the start time of the current feed dispensing cycle through the above process. In the actual dispensing or feeding process of the current feed dispensing cycle, the feed dispensing at the start time of the current feed dispensing cycle is performed at the feed dispensing rate at the start time of the current feed dispensing cycle. The first and second judgment times of the current feed dispensing cycle are also performed at the feed dispensing rate at the start time of the current feed dispensing cycle. That is to say, from the start time of the current feed dispensing cycle to the second judgment time of the current feed dispensing cycle, feed is dispensed to the target feed dispensing point at the feed dispensing rate at the start time of the current feed dispensing cycle obtained above. The judgment time of the current feed dispensing cycle is the time when feed dispensing is paused and the feed dispensing rate is adjusted. The time of the first judgment time of the current feed dispensing cycle is after the start time of the current feed dispensing cycle. Similarly, the time of the second judgment time of the current feed dispensing cycle is after the first judgment time of the current feed dispensing cycle. The time interval between the start time of the current feed dispensing cycle and the first judgment time of the previous feed dispensing cycle is set to be consistent with the time interval between adjacent judgment times. Furthermore, in specific applications, implementers can set the time interval between adjacent judgment moments according to actual conditions such as computing resources, economy, and control precision. For example, if this embodiment requires high control precision, that is, if it requires a high degree of inhibition against insufficient or excessive feeding, then the time interval between adjacent judgment moments can be set to be shorter, such as 1 minute.

[0035] Furthermore, to further avoid problems such as insufficient or excessive feed waste due to changes in the feeding behavior of the farmed animals, this embodiment introduces a feedback control mechanism based on the actual feed consumption, namely, a pause judgment index value. This pause judgment index value reflects the feeding status of the farmed animals at the real-time judgment time of the current feed feeding cycle. Based on the pause judgment index value, it is determined whether to pause feeding and resume the pause, further avoiding problems such as insufficient or excessive feed waste. In this embodiment, the determination of whether to pause feeding and resume the pause is made from the second judgment time of the current feed feeding cycle. This second judgment time of the current feed feeding cycle is recorded as the t-th judgment time of the current feed feeding cycle. Therefore, this embodiment needs to first obtain the pause judgment index value at the t-th judgment time of the current feed feeding cycle at the target feed feeding point. The specific process for obtaining the pause judgment index value at the t-th judgment time of the current feed feeding cycle is as follows:

[0036] First, based on the total cumulative feed weight added to the target feed feeding point at each judgment time from the start time of the current feed feeding cycle to the current feed feeding cycle, and the remaining feed weight in the target feed feeding point at each judgment time, the feed utilization ratio coefficient at different judgment times is obtained. The specific process for obtaining the feed utilization ratio coefficient at the t-th judgment time is as follows: Collect the total cumulative feed weight added to the target feed feeding point at the t-th judgment time from the start time of the current feed feeding cycle to the current feed feeding cycle, and record it as the cumulative feed amount at the t-th judgment time. The weight of the remaining feed in the trough at the target feed delivery point at judgment time t is measured and recorded as the remaining feed amount at judgment time t. The total cumulative feed weight in the trough and the weight of the remaining feed can be collected using relevant weight sensors, etc. The cumulative feed delivery amount at judgment time t is calculated by subtracting the remaining feed amount at judgment time t, and this result is recorded as the cumulative feed intake at judgment time t. The ratio of the cumulative feed intake at judgment time t to the cumulative feed delivery amount at judgment time t is calculated and recorded as the feed utilization ratio coefficient at judgment time t. The formula for the feed utilization ratio coefficient at judgment time t is:

[0037]

[0038] in, Let be the feed usage ratio coefficient at the t-th judgment time. Let be the cumulative feed input at the t-th judgment time. The remaining feed amount at the t-th judgment time; the larger the feed utilization ratio coefficient at the t-th judgment time, the greater the amount of feed consumed by the farmed animals at the target feed placement point from the start of feed placement to the t-th judgment time. Furthermore, the method for obtaining the feed utilization ratio coefficient at any judgment time is the same as the method for obtaining the feed utilization ratio coefficient at the t-th judgment time.

[0039] Then, the preset time length is obtained, and the time period formed by the preset time length before the t-th judgment time and the t-th judgment time is recorded as the local analysis time period of the t-th judgment time. It is required that the local analysis time period of the t-th judgment time belongs to the time period from the start time of feed feeding in the current feed feeding cycle to the t-th judgment time. If the preset time length is greater than or equal to the length of the time period from the start time of feed feeding in the current feed feeding cycle to the t-th judgment time, then this time period from the start time of feed feeding in the current feed feeding cycle to the t-th judgment time is directly used as the local analysis time period of the t-th judgment time, including the t-th judgment time.

[0040] Then, based on the distance between each judgment time and the t-th judgment time in the local analysis time period, the feed usage ratio coefficients at each judgment time in the local analysis time period of the t-th judgment time are weighted and fused to obtain the pause judgment index value at the t-th judgment time. That is, the result of the inverse probability normalization of the distance between the k-th judgment time and the t-th judgment time in the local analysis time period of the t-th judgment time is recorded as the weight value of the k-th judgment time. Inverse probability normalization refers to performing negative correlation mapping and then performing probability normalization. Negative correlation mapping is achieved through a negative exponential function with a constant e as the base. The result of multiplying the weight value of the k-th judgment time with the feed usage ratio coefficients at the k-th judgment times is the weighted feed usage ratio coefficient at the k-th judgment times. The result of summing the weighted feed usage ratio coefficients at all judgment times in the local analysis time period of the t-th judgment time is recorded as the pause judgment index value at the t-th judgment time of the current feed delivery cycle of the target feed delivery point. Furthermore, the smaller the pause judgment index value at the t-th judgment time, the more it indicates that the livestock at the target feed delivery point is no longer feeding or is feeding too little. To avoid overfeeding and waste, feeding should be paused. Conversely, the larger the pause judgment index value at the t-th judgment time, the more it indicates that the livestock at the target feed delivery point is feeding too much. To avoid underfeeding, feeding should continue. However, to further prevent underfeeding or overfeeding, it is necessary to analyze and determine whether the feeding rate needs to be adjusted. The weight value of the k-th judgment time in the local analysis time period of the t-th judgment time is... , Let be the weight value of the k-th judgment time within the local analysis time period of the t-th judgment time. K1 represents the time interval between the k-th and t-th judgment times within the local analysis time interval of the t-th judgment time, and K1 represents the total number of judgment times within the local analysis time interval of the t-th judgment time.

[0041] Therefore, this embodiment can obtain the pause judgment index value at the t-th judgment time through the above process, and the method for obtaining the pause judgment index value at any judgment time after the t-th judgment time is the same as the method for obtaining the pause judgment index value at the t-th judgment time. Therefore, this embodiment will not describe it again. After obtaining the pause judgment index value at the t-th judgment time, it is determined whether the pause judgment index value at the t-th judgment time is less than the preset coefficient threshold. If it is less, it means that the livestock at the target feed delivery point is no longer feeding or the amount of feed is too small at this time or in the future. In order to avoid overfeeding, it is necessary to pause the delivery first. Pausing delivery means setting the feed delivery rate at the target feed delivery point to 0. That is to say, when it is determined that the pause judgment index value at the t-th judgment time is less than the preset coefficient threshold, the target feed delivery point is controlled to perform delivery pause from the t-th judgment time of the current feed delivery cycle to the (t+1)-th judgment time. In other words, the feed delivery rate at the (t+1)-th judgment time is 0. In practical applications, implementers need to set preset coefficient thresholds based on the range of values ​​for the pause judgment indicator, the tolerance for actual feed recycling or waste, etc. For example, in this embodiment, the preset coefficient threshold can be set to 0.8.

[0042] Step S002: Continue to obtain the pause judgment index value at the (t+1)th judgment time of the current feed delivery cycle, and determine whether the pause judgment index value at the (t+1)th judgment time is less than the preset coefficient threshold. If it is not less than the threshold, restore the feed delivery rate at the (t+2)th judgment time to the feed delivery rate before the delivery pause, and calculate the proportional coefficient trend value at the (t+2)th judgment time. Based on the proportional coefficient trend value at the (t+2)th judgment time, determine whether the delivery rate needs to be adjusted. If adjustment is needed, calculate the feed delivery rate at the (t+3)th judgment time based on the feed delivery rate at the (t+2)th judgment time and the proportional coefficient trend value, and control the target feed delivery point to execute the feed delivery rate at the (t+3)th judgment time from the (t+2)th judgment time to the (t+3)th judgment time. If no adjustment is needed, use the feed delivery rate at the (t+2)th judgment time as the feed delivery rate at the (t+3)th judgment time, and continue to obtain the pause judgment index value at the (t+3)th judgment time of the current feed delivery cycle until the delivery end condition is triggered, and end the feed delivery to the target feed delivery point in the current feed delivery cycle.

[0043] In this embodiment, after determining that the pause judgment index value at the t-th judgment time is less than a preset coefficient threshold, and controlling the target feed delivery point to pause delivery from the t-th judgment time of the current feed delivery cycle to the (t+1)-th judgment time, the pause judgment index value at the (t+1)-th judgment time of the current feed delivery cycle of the target feed delivery point is obtained again, and it is further determined whether the pause judgment index value at the (t+1)-th judgment time is less than the preset coefficient threshold. If it is not less than the preset coefficient threshold, it indicates that the livestock at the target feed delivery point has resumed feeding or has a large feeding amount at this time or in the future. To avoid insufficient feeding, feed delivery needs to be resumed first. In other words, if the pause judgment index value at the (t+1)-th judgment time is not less than the preset coefficient threshold, then the (t+2)-th judgment time will be paused. The feed dispensing rate at the judgment time is restored to the rate before the dispensing pause. This means that from judgment time t+1 to judgment time t+2, feed is dispensed at the rate before the pause. The feed dispensing rate before the pause is the same as the rate at the start of dispensing. Therefore, the feed dispensing rate at judgment time t+2 is the same as the rate at the start of dispensing. The trend value of the proportional coefficient at judgment time t+2 is calculated. This trend value reflects the trend of the proportional coefficient at future judgment times and is crucial for determining whether the feed dispensing rate needs adjustment in the future. Adjusting the feed dispensing rate also helps avoid underfeeding. The key measures for addressing overfeeding include calculating the trend value of the proportional coefficient at the (t+2)th judgment time, and determining whether the feed delivery rate needs adjustment based on this trend value. If adjustment is required, the feed delivery rate at the (t+3)th judgment time is calculated based on the feed delivery rate and proportional coefficient trend value at the (t+2)th judgment time. This calculated feed delivery rate at the (t+3)th judgment time is then used to deliver feed to the target feed delivery point. In other words, the target feed delivery point is controlled to receive the calculated feed delivery rate at the (t+3)th judgment time from the (t+2)th judgment time to the (t+3)th judgment time. Between the three judgment times, feed is dispensed at the feed dispensing rate calculated at the (t+3)th judgment time. If it is determined that no feed dispensing rate adjustment is needed, the feed dispensing rate at the (t+2)th judgment time is used as the feed dispensing rate at the (t+3)th judgment time. In other words, if no rate adjustment is determined, feed dispensing continues at the feed dispensing rate at the (t+2)th judgment time from the (t+2)th judgment time to the (t+3)th judgment time. Then, the pause judgment index value at the (t+3)th judgment time of the current feed dispensing cycle is obtained again.It then continues to determine whether the pause judgment index value at the (t+3)th judgment time is less than the preset coefficient threshold. If it is not less, it calculates the proportional coefficient trend value at the (t+3)th judgment time, and based on the proportional coefficient trend value at the (t+3)th judgment time, it continues to determine whether the feed delivery rate needs to be adjusted. If adjustment is needed, it calculates the feed delivery rate at the (t+4)th judgment time based on the feed delivery rate and proportional coefficient trend value at the (t+3)th judgment time, and controls the target feed delivery point to execute the feed delivery rate at that time from the (t+3)th judgment time to the (t+4)th judgment time. If the calculated feed dispensing rate at the (t+4)th judgment time does not require adjustment, then the feed dispensing rate at the (t+3)th judgment time is used as the feed dispensing rate at the (t+4)th judgment time. The target feed dispensing point continues to receive feed at the same rate from the (t+3)th judgment time to the (t+4)th judgment time. The pause judgment index value at the (t+4)th judgment time of the current feed dispensing cycle is continuously acquired until the dispensing termination condition is triggered, at which point the current feed dispensing cycle for the target feed dispensing point is terminated.

[0044] In this embodiment, the conditions for ending the feeding cycle include the cumulative feeding time reaching a preset feeding time and the total weight of feed fed into the trough at the target feed feeding point reaching a preset feeding limit. In other words, the feeding cycle ends when the cumulative feeding time reaches the preset feeding time or the total weight of feed fed into the trough at the target feed feeding point reaches the preset feeding limit. For example, if the preset feeding time has not yet been reached at the y1st judgment time of the current feeding cycle, but the feed fed into the trough at the target feed feeding point has been fed from the start time of the current feeding cycle to the y1st judgment time... If the total cumulative weight of feed added exceeds the preset upper limit, the feeding of the target feed point in the current feeding cycle will end. If the total cumulative weight of feed added to the trough at the target feed point from the start time of the current feeding cycle to the second judgment time (y2) does not exceed the preset upper limit, but the time from the start time of the current feeding cycle to the second judgment time (y2) exceeds the preset feeding time, the feeding of the target feed point in the current feeding cycle must also end. Alternatively, in other real-time methods, simply setting the cumulative feeding time to the preset feeding time can be used as the end condition. In specific applications, real-time users need to set the preset feeding time based on the actual situation, such as the size of the farmed animals and the length of the feeding cycle interval. For example, the preset feeding time can be set to 30 minutes. The preset upper limit can be set by the implementer according to the actual situation. For example, if the total cumulative weight of feed added in a certain historical feeding cycle is the largest, then the total cumulative weight of feed added in that historical feeding cycle can be selected as the preset upper limit.

[0045] In addition, if the pause judgment index value at the t-th judgment time is less than the preset coefficient threshold, and the pause judgment index value at the (t+1)-th judgment time is also less than the preset coefficient threshold, then the feed delivery point is controlled to pause feed delivery until the (t+2)-th judgment time, that is, the feed delivery rate at the (t+1)-th and (t+1)-th judgment times is 0.

[0046] The specific process for obtaining the trend value of the feed use ratio coefficient at the (t+2)th judgment time is as follows: First, obtain the local analysis time period at the (t+2)th judgment time. The method for obtaining the local analysis time period at the (t+2)th judgment time is the same as that for obtaining the local analysis time period at the (t)th judgment time. Based on the rate of change of the feed use ratio coefficient between adjacent judgment times in the local analysis time period at the (t+2)th judgment time and the distance between the judgment time and the (t+2)th judgment time in the local analysis time period at the (t+2)th judgment time, obtain the weighted rate of change of each judgment time in the local analysis time period at the (t+2)th judgment time. The weighted rate of change of the b-th judgment time in the local analysis time period at the (t+2)th judgment time is the result of multiplying the rate of change of the feed use ratio coefficient between the (b+1)-th and b-th judgment times in the local analysis time period at the (t+2)th judgment time by the weight value of the b-th judgment time. The rate of change of the feed use ratio coefficient between the (b+1)-th and b-th judgment times is... Where D is the feed usage ratio coefficient at the (b+1)th judgment time minus the feed usage ratio coefficient at the bth judgment time, T is the time interval between the (b+1)th judgment time and the bth judgment time, the weight value of the bth judgment time is the result of reverse probability normalization of the time interval between the bth judgment time and the (t+2)th judgment time, and the result of summing the weighted change rates of all judgment times in the local analysis time period of the (t+2)th judgment time is recorded as the trend value of the ratio coefficient at the (t+2)th judgment time. The weighted change rate of the last judgment time in the local analysis time period of the (t+2)th judgment time is not included in the summation.

[0047] Furthermore, in practical applications, implementers can determine the preset time length based on actual computational efficiency and the sensitivity and stability of rate adjustment. For example, if a more stable rate adjustment with lower sensitivity is desired, a longer time length can be chosen; otherwise, a shorter time length can be chosen to improve real-time adjustment sensitivity, but this will reduce stability. A trade-off can be made based on actual needs. For instance, the preset time length can be set to 10 seconds, and the pause judgment index value and proportional coefficient trend value can be determined based on the preset time length to avoid randomness at a single moment. Additionally, the process of obtaining the proportional coefficient trend value at other judgment moments is consistent with the process of obtaining the proportional coefficient trend value at the (t+2)th judgment moment.

[0048] The expression for the trend value of the proportional coefficient at the (t+2)th judgment time is:

[0049]

[0050]

[0051] in, Let B be the trend value of the proportional coefficient at the (t+2)th judgment time, and let B be the total number of judgment times in the local analysis time period at the (t+2)th judgment time. This represents the rate of change of the feed usage ratio coefficient between the b-th judgment time and the b+1-th judgment time within the local analysis period of the (t+2)-th judgment time. Let b be the weight value of the b-th judgment time within the local analysis time period of the (t+2)-th judgment time. The time interval between the b-th judgment time and the (t+2)-th judgment time; the rate of change of the feed usage ratio coefficient between adjacent judgment times in the local analysis time period of the (t+2)-th judgment time can reflect the changing trend of the ratio coefficient after the (t+2)-th judgment time. The closer the rate of change is to the (t+2)-th judgment time in time, the better it reflects the changing trend of the feed usage ratio coefficient at the judgment times after the (t+2)-th judgment time. Therefore, in order to make the trend value of the ratio coefficient at the (t+2)-th judgment time more reliable, this embodiment makes the rate of change that is closer to the (t+2)-th judgment time have a higher participation in or a greater dominance over the trend value of the ratio coefficient at the (t+2)-th judgment time. It can reflect the changing trend of the feed usage ratio coefficient at the time of future judgment. A value greater than 0 indicates that the proportion coefficient has an increasing range, a value less than 0 indicates that it has a decreasing range and that supply and demand are unbalanced, and a value equal to 0 indicates that supply and demand are balanced. Supply and demand balance means that the amount of feed given at the target feed delivery point is equal to the amount of feed consumed by the animals at the target feed delivery point. Conversely, supply and demand imbalance means that they are not equal. When they are not equal, problems such as overfeeding or underfeeding are likely to occur, and the feeding rate needs to be adjusted.

[0052] The specific process for determining whether to adjust the feed delivery rate based on the trend value of the proportional coefficient at the (t+2)th judgment time is as follows: If the trend value of the proportional coefficient at the (t+2)th judgment time is equal to 0, it indicates that supply and demand are balanced, and no adjustment to the feed delivery rate is needed at the (t+2)th judgment time, i.e., no adjustment is needed between the (t+2)th and (t+3)th judgment times. If the trend value of the proportional coefficient at the (t+2)th judgment time is not equal to 0, it indicates that supply and demand are unbalanced. To avoid overfeeding or underfeeding, the feed delivery rate at the (t+2)th judgment time needs to be adjusted, i.e., the adjustment is needed between the (t+2)th and (t+3)th judgment times. The method for determining whether to adjust the feed delivery rate based on the trend values ​​of the proportional coefficient at other judgment times is the same as the method for determining whether to adjust the feed delivery rate based on the trend value of the proportional coefficient at the (t+2)th judgment time.

[0053] The specific process for calculating the feed delivery rate at the (t+3)th judgment time based on the feed delivery rate and proportional coefficient trend value at the (t+2)th judgment time is as follows: Determine if the proportional coefficient trend value at the (t+2)th judgment time is greater than 0. If it is greater than 0, it indicates an imbalance between supply and demand, with demand tending to exceed supply. Therefore, to avoid insufficient feeding, the feed delivery rate needs to be increased. Thus, in this embodiment, the feed delivery rate will be... The feed delivery rate is used as the feed delivery rate at the (t+3)th judgment time. If the trend value of the proportional coefficient at the (t+2)th judgment time is less than 0, it also indicates an imbalance between supply and demand, but tends to be more supply than demand. Therefore, to avoid overfeeding, the feed delivery rate needs to be reduced. Therefore, in this embodiment, the feed delivery rate is... The feed delivery rate at the (t+3)th judgment time is defined as follows, where min() is the function for finding the minimum value. Let be the feed delivery rate at the (t+2)th judgment time. sigmoid() is an S-shaped normalization function that maps any real number to the (0,1) interval using an S-shaped curve. This represents the trend value of the proportional coefficient at the (t+2)th judgment time. This is the maximum value within the preset feed delivery rate requirement range. Furthermore, the method for determining the feed delivery rate for the next judgment time after the corresponding judgment time based on the feed delivery rate and proportional coefficient trend value at other judgment times is consistent with the method described above for calculating the feed delivery rate at the (t+3)th judgment time based on the feed delivery rate and proportional coefficient trend value at the (t+2)th judgment time.

[0054] This embodiment of an intelligent feeding control system based on big data aquaculture includes a memory and a processor. The processor executes a computer program stored in the memory to implement the aforementioned intelligent feeding control method based on big data aquaculture.

[0055] Therefore, this embodiment completes the feed dispensing or feeding control at the target feed dispensing point during the current feed dispensing cycle. As another real-time method, the operating status of the feed dispensing device and related sensors is also detected during dispensing. When an abnormality is detected in the dispensing device or sensor data at the target feed dispensing point, feed dispensing at the target feed dispensing point will be stopped, and a safety protection state will be entered. Furthermore, this embodiment adaptively adjusts the feeding rate based on the actual activity status of the farmed animals within the breeding area before feeding, which can minimize the occurrence of localized underfeeding or overfeeding, thereby improving breeding efficiency and feed utilization. In other words, this embodiment, through adaptive adjustment of the feeding strategy, can minimize feed waste and reduced breeding efficiency caused by underfeeding or overfeeding, and can also ensure precise feeding as much as possible. Additionally, in this embodiment, all parameters involved in the formula calculation must be dimensionless, or in other words, all parameters involved in the formula calculation in this embodiment are dimensionless parameters.

[0056] In summary, firstly, the pause judgment index value at the t-th judgment time of the current feed dispensing cycle at the target feed dispensing point is obtained. It is then determined whether the pause judgment index value at the t-th judgment time is less than a preset coefficient threshold. If it is less, the feed dispensing at the target feed dispensing point is paused until the (t+1)-th judgment time. The pause judgment index value at the (t+1)-th judgment time of the current feed dispensing cycle is then obtained again, and it is determined whether the pause judgment index value at the (t+1)-th judgment time is less than a preset coefficient threshold. If it is not less, the feed dispensing rate at the (t+2)-th judgment time is restored to the feed dispensing rate before the pause, and the proportional coefficient trend value at the (t+2)-th judgment time is calculated. Based on the (t+2)-th judgment time... The feed distribution rate trend value is used to determine whether an adjustment is needed. If adjustment is needed, the feed distribution rate at the (t+2)th judgment time is calculated based on the feed distribution rate and the feed distribution rate trend value at the (t+3)th judgment time. The feed distribution rate at the target feed distribution point is then controlled to be the same from the (t+2)th judgment time to the (t+3)th judgment time. If no adjustment is needed, the feed distribution rate at the (t+2)th judgment time is used as the feed distribution rate at the (t+3)th judgment time. The pause judgment index value at the (t+3)th judgment time of the current feed distribution cycle is continuously acquired until the feed distribution end condition is triggered, at which point the feed distribution to the target feed distribution point for the current feed distribution cycle ends. This embodiment uses strategies such as feed distribution pause and feed distribution rate adjustment to control the feeding of the target feed distribution point in the current feed distribution cycle. This can minimize the inflexibility of fixed-point and quantitative feeding methods and minimize local underfeeding or overfeeding, thereby improving breeding efficiency and feed utilization.

[0057] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A smart feeding control method for aquaculture based on big data, characterized in that, The method includes the following steps: Obtain the pause judgment index value at the t-th judgment time of the current feed dispensing cycle at the target feed dispensing point. Determine if the pause judgment index value at the t-th judgment time is less than a preset coefficient threshold. If it is less, control the target feed dispensing point to pause feed dispensing until the (t+1)-th judgment time. Continue to obtain the pause judgment index value at the (t+1)-th judgment time of the current feed dispensing cycle. Determine if the pause judgment index value at the (t+1)-th judgment time is less than a preset coefficient threshold. If it is not less, restore the feed dispensing rate at the (t+2)-th judgment time to the feed dispensing rate before the pause, and calculate the proportional coefficient trend value at the (t+2)-th judgment time. The trend value is used to determine whether the feed delivery rate needs to be adjusted. If adjustment is needed, the feed delivery rate at the (t+2)th judgment time is calculated based on the trend value of the feed delivery rate and the proportional coefficient at the (t+2)th judgment time. The feed delivery rate at the target feed delivery point is then controlled to be the same as that at the (t+3)th judgment time from the (t+2)th judgment time to the (t+3)th judgment time. If no adjustment is needed, the feed delivery rate at the (t+2)th judgment time is used as the feed delivery rate at the (t+3)th judgment time. The pause judgment index value at the (t+3)th judgment time of the current feed delivery cycle is continuously obtained until the feed delivery end condition is triggered, at which point the feed delivery to the target feed delivery point in the current feed delivery cycle is terminated.

2. The intelligent feeding control method for big data-based aquaculture as described in claim 1, characterized in that, The methods for obtaining the pause judgment index value at the t-th judgment time include: Based on the total cumulative weight of feed added to the trough at the target feed point from the start time of the current feed feeding cycle to each judgment time, and the remaining weight of feed in the trough at the target feed point at each judgment time, the feed usage ratio coefficient at different judgment times is obtained. The preset time length before the t-th judgment time and the time period formed by the t-th judgment time are recorded as the local analysis time period of the t-th judgment time. The time interval between each judgment time and the t-th judgment time in the local analysis time period of the t-th judgment time is processed by reverse probability normalization and recorded as the weight value of the corresponding judgment time. The sum of the product of the feed usage ratio coefficient under each judgment time in the local analysis time period of the t-th judgment time and the weight value of the corresponding judgment time is recorded as the pause judgment index value of the t-th judgment time.

3. The intelligent feeding control method for big data-based aquaculture as described in claim 2, characterized in that, The methods for obtaining the feed usage ratio coefficient at the t-th judgment time include: The total cumulative weight of feed added to the trough at the target feed point from the start time of the current feed feeding cycle to the t-th judgment time is recorded as the cumulative feed amount at the t-th judgment time. The weight of feed remaining in the trough at the target feed point at the t-th judgment time is detected and recorded as the remaining feed amount at the t-th judgment time. The ratio of the cumulative feed amount at the t-th judgment time minus the remaining feed amount at the t-th judgment time to the cumulative feed amount at the t-th judgment time is recorded as the feed utilization ratio coefficient at the t-th judgment time.

4. The intelligent feeding control method for big data-based aquaculture as described in claim 2, characterized in that, The method for obtaining the trend value of the proportional coefficient at the (t+2)th judgment time includes: Based on the rate of change of the feed usage ratio coefficient between adjacent judgment times in the local analysis time period of the (t+2)th judgment time and the time interval between the judgment time in the corresponding local analysis time period and the (t+2)th judgment time, the weighted rate of change of each judgment time in the local analysis time period of the (t+2)th judgment time is obtained. The weighted rate of change of the b-th judgment time in the local analysis time period of the (t+2)th judgment time is the result of multiplying the rate of change of the feed usage ratio coefficient between the (b+1)-th and b-th judgment times in the local analysis time period of the (t+2)th judgment time by the weight value of the b-th judgment time. The weight value of the b-th judgment time is the result of reverse probability normalization of the time interval between the b-th and t+2-th judgment times. The sum of the weighted rates of change of all judgment times in the local analysis time period of the (t+2)th judgment time is recorded as the trend value of the ratio coefficient in the (t+2)th judgment time.

5. The intelligent feeding control method for big data-based aquaculture as described in claim 1, characterized in that, The method for determining whether to adjust the delivery rate based on the trend value of the proportional coefficient at the (t+2)th judgment time includes: If the trend value of the proportional coefficient at the (t+2)th judgment time is equal to 0, then it is determined that no adjustment of the feed delivery rate is required at the (t+2)th judgment time. If the trend value of the proportional coefficient at the (t+2)th judgment time is not equal to 0, then it is determined that an adjustment of the feed delivery rate is required at the (t+2)th judgment time.

6. The intelligent feeding control method for big data-based aquaculture as described in claim 1, characterized in that, The method for calculating the feed delivery rate at the (t+3)th judgment time based on the feed delivery rate and proportionality coefficient trend value at the (t+2)th judgment time includes: If the trend value of the proportional coefficient at the (t+2)th judgment time is greater than 0, then... As the feed delivery rate at the (t+3)th judgment time, if the trend value of the proportional coefficient at the (t+2)th judgment time is less than 0, then... The feed delivery rate at the (t+3)th judgment time is defined as follows, where min() is the function for finding the minimum value. Let be the feed delivery rate at the (t+2)th judgment time, and sigmoid() be the sigmoid normalization function. This represents the trend value of the proportional coefficient at the (t+2)th judgment time. This represents the maximum value within the preset feed delivery rate requirement range.

7. The intelligent feeding control method for big data-based aquaculture as described in claim 1, characterized in that, The t-th judgment time is the second judgment time of the current feed feeding cycle. The first and second judgment times of the current feed feeding cycle are based on the feed feeding rate at the start time of feed feeding in the current feed feeding cycle.

8. The intelligent feeding control method for big data-based aquaculture as described in claim 7, characterized in that, Methods for obtaining the feed infeed rate at the start of the current feed infeeding cycle include: During the preset monitoring period before the start of the current feed delivery cycle, acquire videos of the target aquaculture area. The video is divided into frames using a preset grid to obtain region blocks in each frame. In the target aquaculture area, regions that have a physical spatial mapping relationship with the region blocks on the images in the video are obtained and recorded as sub-regions in the target aquaculture area. The result of arranging all the region blocks on all the region blocks on the images in the video that have a physical spatial mapping relationship with each sub-region according to the order of their image acquisition is recorded as the region block sequence of the corresponding sub-region. Based on the number of aquaculture objects in the block sequence of each sub-region and the time interval between the acquisition time of the image to which the block belongs and the start time of feed distribution in the current feed distribution cycle, the activity frequency assessment value corresponding to each sub-region is obtained. Based on the activity frequency assessment values ​​of all sub-regions, the distance between each sub-region and the target feed delivery point, and the preset feed delivery rate requirement range, the feed delivery rate at the start of the current feed delivery cycle is obtained.

9. The intelligent feeding control method for big data-based aquaculture as described in claim 8, characterized in that, The methods for obtaining the activity frequency assessment values ​​for each sub-region include: For any sub-region, a weighted frequent representation value sequence corresponding to the sub-region is obtained. The a-th weighted frequent representation value in the weighted frequent representation value sequence is the result of multiplying the normalized result of the number of farmed objects in the (a+1)-th block of the sub-region's block sequence, the relative change of the number of farmed objects in the (a+1)-th block, and the distance factor of the (a+1)-th block. The relative change of the number of farmed objects in the (a+1)-th block is the ratio of the positive correlation mapping result of the number of farmed objects in the (a+1)-th block to the positive correlation mapping result of the number of farmed objects in the a-th block of the sub-region's block sequence. The distance factor of the (a+1)-th block is the result of inverse probability normalization of the time interval between the acquisition time of the image to which the (a+1)-th block belongs and the start time of the current feed feeding cycle. The result of summing all the data in the weighted frequent representation value sequence corresponding to the sub-region is recorded as the activity frequency evaluation value corresponding to the sub-region.

10. The intelligent feeding control method for big data-based aquaculture as described in claim 7, characterized in that, A method for determining the feed delivery rate at the start of the current feed delivery cycle based on the activity frequency assessment values ​​of all sub-regions, the distance between each sub-region and the target feed delivery point, and the preset feed delivery rate requirement range includes: The result of reverse probability normalization of the distance between each sub-region and the target feed delivery point is recorded as the weight factor of the corresponding sub-region. The product of the activity frequency assessment value of each sub-region and the weight factor of the corresponding sub-region is recorded as the weighted activity frequency assessment value of the corresponding sub-region. The sum of the weighted activity frequency assessment values ​​of all sub-regions is recorded as the initial feed demand assessment value of the target feed delivery point. The result of multiplying the range of the preset feed delivery rate requirement interval with the normalized result of the initial feed demand assessment value and adding the minimum value of the preset feed delivery rate requirement interval is recorded as the feed delivery rate at the start of the current feed delivery cycle.

Images